Home

User Manual

1. variance of the adult density of extant patches demography the above demographic stats for offspring and adults extrate extrate proportion of extinct patches in the population fecundity adlt femfec mean assigned females fecundity adlt femrealfec mean effective females fecundity discounting offspring that do not survive different from the previous one only when viability selection occurs with breeding adlt femvarfec mean variance in effective fecundity of females adlt malrealfec mean effective males fecundity adlt malvarfec mean variance in effective fecundity of males kinship off fsib mean proportion of full sib Table 7 1 continued on next page CHAPTER 7 OUTPUT STATISTICS 86 Stat option Output name Description off phsib mean proportion of paternal half sib off mhsib mean proportion of maternal half sib off nsib mean proportion of non sib off self mean proportion of selfed offspring pedigree ped outb mean proportion of offspring born from an outbred mating between unrelated parents born in different patches ped outw mean proportion of offspring born from an outbred mating between parents born in the same patch but unrelated both parents parents are different ped hsib mean proportion of offspring born from parents with at least one identical parent half sib parents p
2. AaBb element number five in the genotype array The table below shows the ordering of the genotypes in the array BB Bb bb AA 1 2 8 Aa 4 5 6 aa T 8 9 The values should be given relative to maximum fitness 1 Wild type geno types should thus get value 0 and incompatible genotypes should get negative values Otherwise be sure to set the fitness model of the selection LCE to rel ative_local to get relative fitness values see parameter selection_fitness_model dmi_save_genotype bool opt Used to tell Nemo to write the genotypes to file dmi_logtime integer opt Tells every what generation the genotypes should be saved to a text file dmi output dir string opt Tells where relative to root_dir to save the genotype files CHAPTER 5 TRAITS 76 STATS adlt off dmi Records the average frequency of allele 1 the mutant and the average frequency of incompatibility over all loci pairs The output in the stat files include patch specific averages and overall means for both quantities adlt off dmi freq adlt off dmi p The incompatible genotype is AaBb in the diploid case or 01 01 as in the output genotype file and Ab or aB in the haploid case The frequency of these genotypes is recorded in the output stat file as adlt off dmi icmp for the overall average or adlt off icmp p for the per deme frequencies 5 6 Dispersal genes name fdisp mdisp files NA phenotype a real value in 0 1 If the
3. Estimation of fixation indices and gene diversity Ann Hum Genet 47 253 259 Raymond M and F Rousset 1995 GENEPOP version 1 2 population genetics software for exact tests and ecumenicism J Heredity 86 248 249 Weir B S and C C Cockerham 1984 Estimating F Statistics for the analysis of population structure Evolution 38 1358 1370 Weir B S and W G Hill 2002 Estimating F Statistics Annu Rev Genet 36 721 750 5 3 Quantitative traits name quant files quanti output only phenotype continuous value on R Quantitative traits are traits that show a continuous distribution of values also sometimes called metric traits A classic example is body weight a trait that varies continuously both among and within individuals The trait implementation models CHAPTER 5 TRAITS 68 these aspects of trait variation by using a continuum of allele model of mutation where each mutational effects are drawn from a Normal distribution see parameters below In addition a di allelic model is also implemented where mutations can only take two values a This model is provided for comparisons with classical quantitative genetics models The trait architecture is kept simple with additive action of the loci no domi nance no interactions When muiltiple traits are modeled the loci are completely pleiotropic meaning that each locus has an effect on each trait and the mutation effects drawn from a mul
4. and may be different from the previous estimate using Nei and Chesser 1983 That last value will be similar to CHAPTER 5 TRAITS 67 Weir and Cockerham 1984 estimate when sample sizes are equal Note that since version 2 0 8 the Weir and Cockerham 1984 Fsr estimate 0 is also available stat option fstWC Finally the within 0 and between a population coancestry coefficients can also be directly computed using the coa stat options These stats are sometimes referred as kinship or allele sharing coefficients They use the explicit pairwise com parisons of individual sequences to compute the mean population 6 s and between populations a s This method will give exactly the same estimates of the within and between demes Far values using the Weir and Hill 2002 estimates but is more demanding of computer time On the other hand coancestries are given for smaller groups of individuals such as within and between sex or within pedigree classes e g full sib or half sib coancestries etc The Far estimates can be computed from the coancestries as follows ij a l a 0 a 0 Q Fst o gt Pap FE Papi with i and j are population indices with i 4 j These estimates will be equivalents to the Weir and Hill 2002 estimates References Goudet J 1995 FSTAT Version 1 2 A computer program to calculate F statistics Journal of Heredity 86 485 486 Nei M and R K Chesser 1983
5. each genotype The five next lines are the locus names plus the locus names for the four last values the age sex pedigree class and population of origin of each individual This extra information is not processed by the FSTAT program and should thus be removed to be used with that program It is however extremely useful when using this file format to load a new population from a saved simulation file The individuals information will thus be used to assign the individuals to their respective sex and age classes The following lines contain the individual s info one individual per line The first number is the population number in which the individual finds itself at the time of the recording The 5 next numbers columns are the genotype values of each of the 5 loci As in this example we are using two digit per allele the first two digits of a locus genotype number are the first allelic value e g allele 14 for the first allele of the first locus of the first individual while the two next digits are the second allelic value as individuals are diploids here e g allele 14 for the second allele of the first locus of the first individual Each line ends with four numbers The first is the age class 1 offspring 4 adult the second is the sex tag 1 female 0 male the third is the individual s pedigree class that is the pedigree relationship of its parents 0 parents from different demes 1 parents from same deme but
6. 0 or NaN otherwise The size of the dispersal matrix also depends on the number of patches and cannot be automatically updated when specified in input In that case an error message is issued and the simulation is aborted The best workaround is to set the number of patches constant from the start but set the initial carrying capacity of unwanted patch to 0 before adding them at a latter generation by increasing their carrying capacity CHAPTER 2 THE INIT FILE 19 2 7 Output files and naming conventions As briefly explained in the previous section the output files of a simulation have a common base name That name is taken from the argument of the parameter filename see section 3 1 in the init file and any expansion strings are substituted with their corresponding parameter value Several extension strings are then added to that base name Counter extensions A first kind of extension is the generation or replicate num ber or both depending on the periodicity of the output That extension start with an underscore and is followed by a number 002 The number of digit depends on the maximum number of generations or replicates in the simulation For instance if a file is written every replicate and the simulation has 100 replicates the counter will be made of three digits as above The same is true for the generation counter When both counters are added to the filename the generation counter precedes the replicate counter and e
7. 10 19 29 97 100 gt Pop extinction replicate 10 10 10 19 33 100 100 done CPU time 00 00 11s SIMULATION 2 5 replicate 10 10 10 21 00 100 100 done CPU time 00 01 26s SIMULATION 3 5 replicate 10 10 10 24 00 100 100 done CPU time 00 02 55s SIMULATION 4 5 Exe Chapter 2 The input parameters file The configuration file or init file presented previously is a text file with one pa rameter per line in a key value scheme where the key is the parameter name and the value its argument value Each line or string in a line that begins with a is treated as a comment and is ignored Parameters are character strings with no whitespace character that may be followed by one to several argument values sep arated by at least one white space character A particular parameter must appear only once in the init file this is the only restriction for now The order of appearance of the parameters in the file does not matter 2 1 Parameter types Here is a list of the different types of argument a parameter can take e boolean bool works on a presence true absence false basis when no argument is passed Also accepts 1 as true or set and 0 as false or unset this is especially useful for temporal arguments see below e integer argument is a dot less number value a limit to t
8. Table 7 6 Selection stats continued on next page CHAPTER 7 OUTPUT STATISTICS 94 Stat option Output name Description prop outw proportion of w n demes outbreds prop hsib proportion of half sib crossings prop fsib proportion of full sib crossings prop self proportion of selfed progeny survival survival outb mean proportion of surviving offspring after viability selection for each pedigree class survival outw survival hsib survival fsib survival self off ad1t fitness patch age W avg p1 mean offspring adult fitness of patch 7 loff adlt fitness var patch Table 7 6 Selection stat options continued age W var p1 mean offspring adult variance in fitness of patch i 7 9 Dispersal Table 7 7 Dispersal stat options Stat option Output name Description adlt off disp age disp mean dispersal rate age fdisp mean female dispersal rate age mdisp mean male dispersal rate Table 7 7 Dispersal stat options continued 7 10 Wolbachia Table 7 8 Wolbachia stat options Stat option Output name Description wolbachia off fwoinf mean infection frequency of offspring females Table 7 8 continued on next page CHAPTER 7 OUTPUT STATISTICS 95 Stat option Output name off mwoinf Description mean infection frequency of offspring males off incmating mean number of inc
9. The markers implemented here are all diploid nuclear markers Two models of mutation are implemented the SSM Single Step Mutation and the KAM K Allele Model models see below for details The probability of crossing over occurrences between two adjacent loci can be set by the parameters of the genetic map The number of alleles and the allelic mutation rate are constant across loci New populations can be initiated by assigning random allelic values within the range 1 ntrl_all to each locus thus assuring a very large initial variance or by assigning the same value to all loci Other initialisation options are given by the source_pop option above see population parameters 3 2 which allows you to load a population s genotypes from an FSTAT input file see below for a description of that file format or with the ntrl_init LCE to specify patch specific allele frequencies for di allelic loci see section 4 9 2 ntrl loci integer Number of diploid neutral markers per individual ntrl_all 1 to 256 Number of alleles per neutral locus same number for each locus ntrl_mutation_rate decimal Mutation rate of the neutral alleles identical across loci T he mutation model is specified with the next parameter ntrl mutation_model 0 1 2 Available mutation models are 0 no mutations 1 SSM Single Step Mutation 2 KAM K Allele Model The no mutation model 0 is simply a void model used for the case of a null muta
10. any platform supporting a console like environment and allowing it to be compiled with standard C C compilers GNU gcc being the default Installing Installing Nemo is straightforward you just need to copy the binary file corresponding to your operating system from the hosting web site http nemo2 sourceforge net and use it at once or in the case your operating system is not supported copy the source code compile it and use the executable See CHAPTER 1 INTRODUCTION 2 the documentation provided with the source package for instructions concerning the compiling process Using The basic users interface is a text file a k a the init file containing the input parameters and their argument in a key value scheme Nemo is then launched from the console with that init file as an argument Some runtime infor mation current running simulation current generation replicate etc is written to the standard output terminal window Nemo also gives the possibility to save the simulation data to a variety of files in text or binary format depending on the options chosen in input The user may save the traits complete genotypic informa tion the simulation s summary statistics or the complete state of the population periodically See chapter 2 for input directions and chapter 3 for parameters de scription Extending Nemo is designed as a flexible and extensible coding framework It is aimed at facilitating the implementatio
11. arguments is updated before the first event in the life cycle Temporal argument string must always start with the initial argument value specified as g0 and arguments are separated by commas parami g0 valuel g100 value2 g10000 value3 This example specifies three different parameter values that will be used throughout the simulation value is used at initialization of the simulation and beginning of each replicate value2 and value3 are used at generation 100 and 10 000 respectively The component that declares and uses param1 will update itself at the specified generations Temporal parameters can thus be used to dynamically modify the state of the population through time to model population fragmentation or bottlenecks for instance The following example shows how to progressively fragment a population while keeping its total size and number of migrants constant at 10 000 and 1 respectively patch_number 960 10 g5000 15 g10000 20 patch_capacity 990 1000 g5000 666 g10000 500 dispersal_rate g0 0 001 g5000 0 0015 g10000 0 002 Important Note Changing the number of patches during a simulation can lead to various problems at runtime as many features depend on it For instance the number of patch specific stats cannot be updated this would cause a lot of mess in the stat output files and thus data will not be recorded for the added patches they will be set to
12. as the number of combinations of the sequential argument values present in the configuration file Each simulation receives a different output filename that might be explicitly defined in the configuration file or automatically generated This section explains how to specify specific simulation output filenames CHAPTER 2 THE INIT FILE 15 based on the sequential parameter values This mechanism also works throughout the whole set of string parameter arguments e g the output directory or input binary file arguments Basic filename argument string expansion If your configuration file com prises sequential parameters you may add the special expansion character followed by a number 1 for e g in the base filename argument string to build specific file names for each simulation initiated by the sequential parameters see description of the filename parameter in section 3 1 This expansion character can also be used in any string argument of any simulation parameter throughout the init file and will be expanded in the exact same way as for the base filename The number after the expansion character refers to a specific sequential parameter present in the init file starting with 1 for the first The sequential parameters are alphabetically sorted so that the number one is not the first in the file but the first in alphabetical order You cannot use more expansion characters than the number of sequential parameters but if you use less or none a
13. bp the effective fecundity of between deme matings mating partners are from different patches Table 7 4 Deleterious mutations continued on next page CHAPTER 7 OUTPUT STATISTICS 92 Stat option Output name age load Description mean demic mutational load computed as L 1 ma where Wmax is the maximum number of surviving offspring produced by a female in a patch Note heterosis and load are computed from the female fecundities which are updated according to the offspring survival in the breed_selection LCE only and are thus null when viability selection is performed differently In that case they can be inferred from the fitness stats adlt off viability age viab mean patch viability mean trait value age viab outb mean viability of outbred individuals between demes age viab outw mean viability of outbred individuals within demes age viab hsibs mean viability of inbred individuals between half sib parents age viab fsibs mean viability of inbred individuals between full sib parents age viab self mean viability of inbred individuals descended from selfed parent age prop outb proportion of between demes outcrosses age prop outw proportion of within demes outcrosses age prop hsibs proportion of within demes half sib matings age prop fsibs proportion of within demes full sib matings age prop self pro
14. by default stat_output_no_means bool opt Suppresses the writing of the output file containing the stat means ending with _bygen txt Output stats alive rpl This stat appears in the _bygen txt files only and is the number of alive replicateat each generation recorded This is an automatic statistic no addi tional token is needed to the stat parameter CHAPTER 4 LIFE CYCLE EVENTS 93 4 14 Saving Files name save files integer age flags unchanged files varies This LCE tells the program when during the life cycle the simulation data must be saved on disk by the different simulation components This excludes binary data that is saved by the store LCE see below The save files LCE is mandatory if you want to have any output data saved by your simulation Each simulation component trait or LCE may define different output files to save specific information e g specific stats or genotypes phenotypes of a specific trait etc The program file manager is notified by save files that is must initiate the file handlers output process at the point it has been inserted in the life cycle The type composition of the data that is saved will thus depend on the rank of this LCE in the life cycle because the age composition and the state of the population is changed by other LCEs It is not possible for now to use save_files more than once in the life cycle This prevents for instance saving some data before and after a spe
15. demes to fill the receiving demes to their carrying capacity The receiving population is filled sequentially starting from the first patch to the last No extra demes are added to match the number of demes in the source population The population structure will be perfectly preserved if the source and the target population have same deme structure and sizes If not present then the individuals are randomly sampled from the source population without replacement In that case the individuals of the source population are gathered together in a single container from which they are sampled until the whole receiving population is full or the source population is empty source fill age_class adults offspring opt This sets the age class to load from the source population It overrides the rule described above using the required age class of the life cycle events of the current life cycle CHAPTER 3 SIMULATION COMPONENTS 27 source generation integer opt The generation to load from the binary source file The population initializa tion will fail if that generation is not present in the binary file The binary files may indeed store more than one generation see section 4 15 source _replicates integer opt By specifying this parameter you can tell Nemo how many replicates of the source population have to be used throughout the simulation to load the pop ulation from If the value given here matches the number of replicates
16. dominance coefficients are set for a given locus and apply to all individuals within the species The total fitness of an individual depends on the way the mutations interact and two fitness models are available a multiplicative fitness model independent action of the different mutations the default and an additive fitness model non independence among loci delet_loci integer Number of deleterious loci per individual The initial mutation frequency can be set below By default the initial genotype is all wild type delet_mutation_rate decimal Deleterious mutation rate allelic mutation rate from the wild type to the deleterious form only There is no reverse mutation rate for now delet_mutation model 1 2 opt There are two different models of mutation 1 default the location of each new mutation is randomly drawn irrespec tive of the presence of a mutation at that location 2 the location of a new mutation is redrawn each time it appears at a ho mozygous deleterious locus delet_recombination_rate delet_genetic_map delet_random_genetic_map Recombination is handled by the genetic map All genetic map parameters apply See section 5 1 delet_init_freq decimal opt Initial allele frequency of the deleterious allele If the parameter is absent the initial number of mutations of each individual is null The initial mutations are randomly placed number initial frequency times the number of locus CHAPTE
17. following parameters are added to the init file two quantitative traits will be added to the individuals One codes for the female dispersal rate and is expressed in females The second codes for the male dispersal rate and is expressed in males only Both traits are continuous quantitative traits coded by a single diploid locus whose allele values are real numbers ranging from 0 to 1 The two loci are co inherited The dispersal probability of an individual i e the trait s phenotype is the mean of the two allele values at the corresponding locus disp_mutation_rate decimal Mutation rate of the dispersal alleles This is the probability to change the allele value by an amount drawn from an inverse exponential distribution with the mean set below disp_mutation_mean integer This parameter is the mean of the exponential distribution used to draw the mutation step added to the genotype value disp_init_rate_fem decimal opt Initial genotype both alleles of the female dispersal locus disp_init_rate_mal decimal opt Initial genotype both alleles of the male dispersal locus disp_init_rate decimal opt Initial genotype of both the male and female dispersal locus CHAPTER 5 TRAITS TT 5 7 Wolbachia name wolb files NA phenotype a boolean representing the infection status of the individual The Wolbachia trait is used to simulate the dynamics of an endosymbiotic parasite causing cytoplasmic incompatibility Its tran
18. infected with Wolbachia 4 4 Dispersal name disperse integer age flags offspring required files NA derived components disperse_evoldisp breed_disperse breed_selection_disperse Moves offspring among patches according to the migration scheme chosen Dispersal rates are taken as forward migration rates that is they represent the probability of an individual to move from patch to patch j These rates will be equivalent to immigration rates under the classical models of island model migration and stepping stone migration Forward migration is equivalent to zygotic diploid migration as opposed to backward migration modelled by the breed_disperse LCE as gametic haploid migration There are three mutually exclusive ways of specifying the migration rates in Nemo 1 by specifying a sex specific dispersal rate and migration model e g Island Model Stepping Stone model etc ii by specifying the full migration matrix allowing for more flexibility in the type of migration modelled e g allowing for long distance dispersal on a landscape iii new in 2 3 by specifying the reduced migration matrices which holds the non zero migration rates only and allows the modelling of large landscapes with sparse dispersal matrices This last option is an optimisation for modelling large grids with limited dispersal among patches and brings a large speed up compared to the previous implementations All migration matrices are now r
19. is the default 4 12 Population Regulation name regulation integer age flags adults required files NA Population regulation is used to remove all individuals in excess of the sex specific carrying capacity of each patch The mode of regulation is therefore called ceiling regulation Regulation is performed on each age class present in the population that is on the offspring and adult individuals for now The supernumerary individuals that are removed are chosen at random It is not necessary to place regulation after aging in the stack of life cycle events as the aging LCE also performs regulation The patches will be at their carrying capacity only if there was enough individuals present prior to regulation 4 13 Save Stats name save_stats integer age flags unchanged files txt _bygen txt This LCE is used to tell the stat services of the simulation to record the summary statistics specified with the stat parameters see below The statistics recorded depend on the age state of the population The position of this LCE in the life cycle is thus important Putting it after breeding will allow you to record stats on both offspring and adults while putting it after aging will allow you to record the stats on the adults only The recorded stats are dumped to a text file at the end of each replicates and at the end of a simulation for the averaged stats but only if the save_files LCE is present in the life cycle See
20. life cycle events to run a basic simulation Note that you can also use Nemo to simply load a previously saved population from a binary file see the source pop population parameter below and compute statistics on it or extract genotypes and save them in a human understandable format usually text Each component and its list of parameters are presented here Some parameters are mandatory they must be present in the init file in order to include a component to a simulation Each component has at least one mandatory parameter Optional parameters are marked as opt below and are used to add extra features needed to build a particular model Nemo will not complain if a mandatory parameter is missing for a non mandatory component i e others than the simulation and population components so you have to be careful while building the init file The parameter type is given between two enclosing square brackets see chapter 2 for details about the different types of parameters There are two main types of simulation components the Traits chapter 5 and the Life Cycle Events chapter 4 The traits are carried by the individuals in the population while the LCEs act as modifiers of the population state and hence act on the individuals state as well as defined by their traits state The action of an LCE may depend on the values of the individual s traits or not For instance selection will remove individuals by checking the phenotype
21. locus positions in units of recombination the centimorgan cM This means that bi allelic neutral sites can be placed on the same map as QTL and sites under background selection the deleterious mutations where loci with the same map position will be physically linked The traits available in Nemo are distinguished by the interpretation of their phenotypic value and their genetic architecture At the coding level this means that different data structures can be mixed together making trait implementation highly flexible and dependent on the specific need for the different traits For instance the neutral ntrl trait has no phenotypic value and is coded on one byte per locus the deleterious delet trait has fitness as its phenotype and is coded on one bit per locus along with a single table relating mutation to their fitness value while the quantitative quant trait has a continuous value as its phenotype coded on one double precision number 8 bytes per locus per trait all loci are pleiotropic when more than one trait is modeled The data structure chosen obviously conditions the number of alleles available per locus hence the use of bits for a bi allelic trait like deleterious mutations and the use of a single byte for neutral markers that can have from two SNP to four nucleotide to many alleles but not an indefinite number of alleles maximum is 256 allelic states Simulation of large DNA sequences Nemo has not been developed to
22. matrix patch 7 adlt off age qevectii pj quanti eigenvect1 patch loadings of trait 7 on the first eigenvector of the G matrix in patch 7 Table 7 3 Quantitative traits continued on next page CHAPTER 7 OUTPUT STATISTICS 91 Stat option adlt off quanti skew patch Output name age Sk q pj Description skew of the phenotypic distribution of trait 7 in patch 7 lad1t off quanti patch adds the stats from quanti mean patch quanti var patch quanti covar patch and quanti eigen patch Table 7 3 Quantitative traits stat options continued 7 6 Deleterious mutations Table 7 4 Deleterious mutations stat options Stat option Output name Description adlt off delet age delfreq mean deleterious mutation frequency age delhmz mean deleterious mutation homozygosity age delhtz mean deleterious mutation heterozygosity age delfix mean number of fixed mutation in the whole population age delfixp mean demic number of fixed mutation age delsegr mean number of segregating mutation in the whole population age delsegrp mean demic number of segregating mutation age delfst Fst of the deleterious mutations age lethequ mean number of lethal equivalents age heterosis heterosis computed as H 1 z b the effective fecundity of within deme matings mating partners are from the same patch
23. model evolution of genetic polymorphisms of DNA sequences on large chromosomal regions spanning several million base pairs at the nucleotide level The reason is that Nemo uses an explicit representation of each locus in each individual This straightforward implementation is fine when mod elling limited number of loci in the range of 100 to 10 000 loci especially for neutral markers The implementation capitalises on position ordered arrays of loci which makes access to locus values an efficient constant time operation This is of par ticular interest for non neutral traits which individual values must be read in each individual at each generation to determine fitness The approach however has a huge computational cost when modeling large sequences of over 10 loci in large populations because of the intense memory usage it entails CHAPTER 1 INTRODUCTION 7 1 2 4 Statistics and outputs Nemo provides several ways to record the ancestral population states Summary statistics can be computed at different time periods during a simulation The statis tics recorded depend on the simulation components used Each simulation compo nent can define its set of statistics that the user can choose among to monitor during a simulation Here are examples of the summary statistics e Neutral trait stats Heterozygosities F stats Far Ggr and 0 Fis Fir allele numbers number of fixed alleles per locus coancestries Nei s D genetic distance et
24. monoecious organisms selfing fusion of self gametes cloning no meiosis suppresses recombination e Dispersal models forward zygotic and backward gametic migration can be modelled with the following dispersal models sex specific dispersal matrices fully describing any complex dispersal pat terns as defined by the user separate seed and pollen dispersal matrices for monoecious organisms large migration matrices for simulations on large geographical grids can be simplified and passed as reduced dispersal and connectivity matrices or pre defined dispersal models with Island Model with migrant or propagule pool migration Stepping Stone Model nearest neighbour migration on a string of patches 2D lattice model on a grid set as a torus or with reflective or absorbing borders e Traits Universally deleterious mutations di allelelic mutations affecting fitness Neutral markers from SNPs to microsatellites Pleiotropic quantitative trait loci multiple correlated phenotypic traits Bateson Dobzhansky Muller incompatibility loci pairs of epistatic loci Dispersal quantitative loci male and female specific dispersal genes Wolbachia endosymbiotic parasite causing cytoplasmic incompatibility CHAPTER 1 INTRODUCTION 5 1 2 1 Population models Besides its flexibility in the types and number of components included in a simula tion Nemo provides a highly versatile populatio
25. name expansion used in the character string of an argument to insert the value of another parameter when that parameter has multiple argument values see sequential parameters in section 2 4 e external parameter file Ofilename used to pass an argument value to a parameter when that argument value e g a large matrix is contained in a separate file The character string filename contains the path to that separate file containing the argument value s see section 2 5 e specifiers g this short character string is used to specify the generation at which a temporal argument value applies For instance Og100 designates a temporal argument value that will be used at generation 100 see section 2 6 Specifiers must be found within a block argument see below e block argument argi arg2 argument values enclosed with two parentheses are treated in a special way Parentheses are used when several arguments and their specifiers must be passed to a parameter without being interpreted as a sequence Such a case appears when specifying temporal argument values see section 2 6 Argument values are separated by commas within a block argument e g g0 0 02 g5 0 5 CHAPTER 2 THE INIT FILE 14 2 3 Matrix parameters A matrix argument may be passed to a parameter in the init file This type of argument contains integer or floating point values separated by commas and curled brackets Here is an example patch
26. population patch_nbfem patch_nbmal integer matrix opt The number of males or females per patch can be given separately with these two optional parameters Each or both of them can be a matrix parameter giving the sex specific sizes of each patches If one of the two sex specific size parameters is missing population initialization will abort Examples The following setting will build a population of 5 patches of different sizes but with equal sex ratio in each patch patch_capacity 10 4 18 20 24 This parameter is sufficient to build a population as the size of the vector will tell the number of patches present In this other example however the number of patches must be given explicitly as no matrix arguments are present patch_number 5 patch_nbfem 8 patch_nbmal 4 This other example will also work fine patch_nbfem 5 patch_nbmal 4 4 3 3 1 Note however that the following will issue a fatal error patch_capacity 10 patch_nbmal 4 4 3 3 1 Indeed patch_capacity has precedence over patch_nbmal and in that case patch_number is missing The correct form would be patch_nbfem 6 6 7 7 9 patch_nbmal 4 4 3 3 1 CHAPTER 3 SIMULATION COMPONENTS 25 This also means that including both patch_capacity and the sex specific size parameters will cause Nemo to ignore the later and use the first one only to build the population
27. replicate counter string may be automatically added when multiple replicate sources are used see below The file extension may also be specified below in the case the file format differs from the default binary one i e bin Only one file can be used for a given simulation or replicate If every replicate of the current simulation is going to use the same binary source file as specified here the full name of the file must be specified 1 0 including counters and extension strings If the source population is to change during a simulation the parameter source replicates must be specified see below In this case the path given here must not be terminated by the usual replicate counter string and any file extension source file type string opt The argument here is the file extension string of the source file including the dot e g bin dat etc This will determine the file format of the source data The default value is bin source_preserve bool opt With this parameter the deme structure of the source population is preserved This means that Nemo will copy individuals from the source population into the current population deme by deme If the source population has less demes than the receiving population then the receiving population will have empty demes Similarly in preserve mode the receiving population will not be full if the source population does not contain enough individuals within
28. the number of lethal equivalents present in the population direct default The fitness of the individual is directly given by the phe notype of the trait as for the deleterious mutations trait This is the default model gaussian Stabilising selection on a set of quantitative traits The fitness of an individual with phenotypic values z is W 2 exp 5 2 0 w z 0 where is a vector of local optimal trait values and w is the variance covariance matrix of selection describing the individual fitness surface quadratic A quadratic model of stabilising selection on a single quantita tive trait Individual fitness is given as Zik Ok 2 E Wis 1 zik a where Zi is the phenotypic value of individual in patch k 6 is the phenotypic optimum in patch k and wy is the inverse of the strength of selection on the trait in patch k Parameter selection_local_optima specifies the values for the 6 s and parameter selection_variance the values for w2 selection_fitness_model absolute relative_local relative_global opt This sets how the fitness of the individual is interpreted By default the fitness of the trait is taken as absolute it does not depend on the fitness of the other individuals in the population Alternatively the fitness of an individual or its survival probability can be interpreted relative to the mean fitness of other individuals in its patch option relative_local or in the w
29. unrelated 2 parents are half sib 3 parents are full sib and 4 selfed mating and the last one is the identifier of the population where that individual was born This file format is close to the FSTAT input file format see J r me Goudet s soft ware http www2 unil ch popgen softwares fstat htm with the addition of the four last columns of the individual data The HIERFSTAT R package see http ww2 unil ch popgen softwares hierfstat html by the same author provides R routines called read fstat data to extract data from an FSTAT file within the R software http www r project org Note about the statistics Nemo lets the user choose between various estimates of gene diversity and genetic differentiation both within and between populations The classical F statistics are available by using the fstat stat option see sec tion 7 2 for more details This option will give the estimates of heterozygosities Ho Hs and Hr and of F statistics Fig Far and Fyr using the weighting method of Nei and Chesser 1983 for unbiased estimates when population sizes vary Another set of F statistics is given by the weighted fst stat options that use the Weir and Hill 2002 unbiased estimates of within and between populations Fsr s for varying sample sizes These stat options may be used to output the whole population matrix of pairwise Far values within and between populations The mean total population weighted Far is also given
30. 1 This trick helps boost the simulations when the starting conditions for the traits under selection are very far from their optimum breed_selection_disperse fitness_threshold decimal opt The minimum fitness value used to rescale the individuals fitness when the mean patch fitness is too low to allow for the patch to be filled see above It is 0 05 by default 5 surviving probability CHAPTER 4 LIFE CYCLE EVENTS 99 Note for version 2 3 since breed_selection_disperse inherits parameter definitions from breed_disperse the dispersal parameters must also use the breed_disperse prefix instead of dispersal see section 4 18 above Chapter 5 Traits The traits described here are e ntrl neutral markers including microsatellites SNPs etc quant quantitative traits delet deleterious mutations dmi Dobzhansky Muller Incompatibility loci fdisp mdisp sex specific dispersal wolb Wolbachia endosymbiotic parasites Each trait has an identifying name or type and may define different output files and stat options For a complete description of the stat options have a look at chapter 7 5 1 The Genetic map New in version 2 3 The three sequence based traits ntrl quant and delet share a common genetic map on which the loci of the different traits are placed The genetic map in Nemo is a recombination map where the locus positions are specified in centi Morgan cM in opposition to the base pair unit bp
31. 3 2 1 Loading a population from a file This section describes the set of parameters needed to load read a population from a file The type of data that can be loaded depends on the file format The binary files written by the store component see 4 15 store the whole population state that is all individuals in the population are saved with their attributes and traits data i e genetic data Other simulation components may define an input function for the type of data they handle Typically the neutral markers trait section 5 2 saves and loads its data in the FSTAT format text and the deleterious mutation trait section 5 4 saves and loads from a text file see respective trait s description for details about those files Filling the population The population loaded is used to set the starting gener ation of a replicate Each replicate may start from a different source replicate file or from a single source file see below The default loading mode randomly draws individuals from that source population without replacement to fill the current pop ulation The two populations may thus have different sizes but it is a good idea to have a source population that is at least as big as the receiving one to completely fill the first generation Unless the source population is loaded in preserve mode see below the structure of the source population is not preserved all individuals in the different patches are pooled together Filling ag
32. 8 dispersal connectivity_matrix matrix opt This matrix specifies to which patch each focal patch row wise is connected through migration The number of elements per row can vary among rows but must be exactly the same as in the dispersal_reduced_matrix It is advised to sort the connected patches in descending order of the migration probability Note At least one of the optional dispersal rate matrix parameters above must be present in order to correctly set the disperse LCE 4 5 Seed dispersal name seed_disperse integer age flags offspring required files NA This LCE is an alias for the disperse LCE as just described above It is used when two types of dispersal events are part of the life cycle as for instance when pollen dispersal i e backward gametic migration is modelled using the breed_disperse LCE The seed_disperse LCE is thus adequate to model zygotic forward migration All parameters are identical to the disperse LCE to the exception that the dispersal prefix must be replaced with seed_disp e g dispersal_rate becomes seed_disp_rate 4 6 Evolving Dispersal name disperse_evoldisp integer age flags offspring required files NA inherits from disperse This is a specialization of the previous LCE and thus inherits its parameters though the rate parameters have no meaning here In addition it defines a couple more parameters used by the evolving dispersal models dispersal_c
33. ARS S 1 3 1 Launching Nemo from the command line 1 3 1 1 For Linux and Mac OS X users 1312 Por Windows USO saci s Wace haw be hae 4 132 Bath III 2 THE INIT FILE Mo III 23 Special characters ic a RE AE wd C Matriz paramelerg se scoe toe core A A EES oi 24 Sequential parameters lt 2 6 2 de be eee re e toas 2 5 External argument files 2 6 Temporal arguments lt lt 26220 Bad ed A eR we ew 2 7 Output files and naming conventions 3 SIMULATION COMPONENTS 3 1 Simulation 2 a a 0000002 be ee 10 12 12 13 14 14 17 18 19 21 CONTENTS Oo N TT ira da AAA 3 2 1 Loading a population from a files 2 2 6 6 es saw ws 4 LIFE CYCLE EVENTS Ll ARDD gc Se hea d tee ee ee eee bee eb ORS ee AREA da Breeding AAA 43 Breeding with Wolbachia c ase c es g e t weed a eae daw wb LA pad oe cocs eiee a A a A i do DRAPER ee AG Evolving Dispersal aires Giu rng SA CE e a Kee ee ot a esc S RTE ee 4 7 1 Multi trait selection gt gt os c o RR 4 7 2 Fixed selection model parameters 4 7 3 Gaussian and quadratic model parameters 4 8 Extinction and Harvesting aa aD Trait initialization oo ee ia AA 4 9 1 Initialization of trait quant 2er 202 Intiallization of trait fil ooe ee ee ered op eee s 49 3 Initialization of trat dMi lt s sechas tagtha ragiu i LIO Resize Populatio
34. E And the cycle starts again The Life Cycle Events described here are aging increase the age of the individuals perform patch regulation breed mate and breed create new offspring generation breed wolbachia breed and Wolbachia transmission infection breed disperse breed with backward migration Wright Fisher model breed_selection breed with selection faster breed_selection_disperse all in one Wright Fisher with selection cross perform a half sib full sib mating design NCI disperse offspring dispersal disperse_evoldisp offspring dispersal with evolving dispersal rates extinction random patch extinction or harvesting regulation patch regulation to carrying capacity resize modify population size patch number and or size save_files write output files to disk save stats record statistics selection perform viability selection on the offspring generation store save simulation data to binary files The LCEs often act as modifiers of the population state Most of the time this simply consists of changing the content of various individual containers either by moving individuals between them or by adding removing individuals to from them Individual containers are ordered by age class and by sex and are aggregated within patches The two main age classes are the adult and the offspring age classes A particular LCE will in general be associated with one or more age class This information is given below by the age f
35. NEMO release 2 3 version 2 3 44 User Manual August 3 2015 authors Fr d ric Guillaume guillaum zoology ubc ca Jacques Rougemont MPI version Jacques rougemont isb sib ch contributors Samuel Neuenschwander Alistair Blachford Sam Yeaman availability http sourceforge net projects nemo2 2006 2015 Fr d ric Guillaume Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying provided also that the sections entitled Copying and GNU General Public License are included exactly as in the original and provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one Permission is granted to copy and distribute translations of this manual into another language under the above conditions for modified versions except that this permission notice may be stated in a translation approved by the Free Software Foundation Contents 1 INTRODUCTION E AI T ri EA A ee ke ew LA MOUSSE ER Se OR oe ed eee ee e 1 2 1 Population models ne ee en ee eee eee een lila Tie Teva z o ee nido ee EEE rs er ETE ck ee ee he hee be daw dc eee Rede Loe Statistics and ONO ce o s seci SE oe ew G L3 Using Nemo gt o c e baw acerte ethane eed 44 ee RE
36. R 5 TRAITS 72 delet_effects_distribution constant exponential gamma lognormal opt The mutational effects can either be a constant value across all loci default option or follow a distribution as set by this parameter Possible distributions of effects are the exponential gamma and log normal distributions The mean effect size and the shape of the distribution are set by the parameter below The dominance coefficient also follows a distribution and is scaled to the mu tational effects using the following relationship h exp ks 2 where k is a scaling factor chosen so that the average dominance coefficient of all mutants is equal to h i e k log 2h s and 3 is the mean effect size constant default all loci have same selection and dominance coefficients This is the default if not specified exponential mutational effects follow a reverse exponential distribution The mean of the distribution is taken from parameter delet_effects_mean gamma the gamma distribution takes two extra parameters beside the mean effect The first is the shape delet_effects_dist_param1 and the second is the scale delet_effects_dist_param2 of the distribution Only the shape is mandatory The scale can be deduced from the mean and shape param eter values mean scale shape lognormal the log normal distribution is another leptokurtic distribution with two mandatory extra parameters js and g the mean and standard deviation o
37. The mutation rate identical for all loci The mutation effect s depends on the allelic model set below quanti_allele_model diallelic diallelic_HC continuous continu ous_HC opt Two ways to model the mutational effects diallelic if mutations can only take a given value or two different values see below or continuous if mu tations are drawn from a Normal distribution with variance and correlation for the multiple traits set below The default model is continuous CHAPTER 5 TRAITS 69 The two Hous of Cards HC variants specify a different way of modelling mutations In the non HC models a new mutation effect is added to the existing allelic value whereas in the HC models the new effect replaces the existing allele quanti_allele_value double matrix opt The effect size of the mutation s or allelic values at a loci in the di allelic mutation model If a single value is given that value is used for all loci A matrix can be used to pass locus specific values If the matrix has a single row an array the mutational effects are the given values at each locus Two different values per locus can be specified if two rows are provided instead of one The number of columns of the matrix must match the number of loci quanti_mutation_variance double opt The variance of the Normal distribution of the mutational effects the mutation effect size in the continuous mut
38. _capacity 1120 20 5 10 5 dispersal_matrix 10 2 0 0 0 0 0 4 0 4 0 4 0 2 0 0 0 0 0 4 0 4 0 4 0 2 0 0 0 0 0 0 0 4 0 4 0 2 0 0 0 0 0 0 0 4 0 4 0 2 lt is mandatory 0 4 0 0 0 0 0 3 0 3 10 3 0 4 0 0 0 0 0 3 10 3 0 3 0 4 0 0 0 0 10 0 0 3 0 3 0 4 0 0 10 0 0 0 0 3 0 3 0 4 The matrix is enclosed by two external brackets within which each row is specified by two internal enclosing brackets 4 Inside a row the column values are separated by commas or semi colons The rows can be separated by any kind of characters but a backslash A matrix argument can as well be used to pass only an array of values as in the first example above or a complete matrix Several matrices may be passed as arguments to a parameter That parameter will then become a sequential parameter see below The different matrices must start on the same line to be sequential arguments The line continuation character V is mandatory if one wants to split matrices over several lines see example above Note that the lines within a matrix do not count the rows can be written over several lines without using the line continuation character 2 4 Sequential parameters A parameter with several argument values on a single line is called a sequential parameter in the sense that it will initiate a sequence of simulations There will be as many simulations
39. ach start with an underscore like this mysim_1000_01 mysim_2000_01 mysim_5000_10 This way the simulation can save each generation for each replicate in a different file The behavior of the various output files i e their periodicity depends on the kind of data the simulation will generate which depends on the user s defined parameters Typically trait genotype files are written per generation and per replicate while binary output files are per replicate only Type extension The second kind of extension string is the file type e g txt and is a classical extension starting with a dot followed by a few characters added to the end of the file name Nemo generates a few basic output files with different types These are the log these files are automatically generated in every folder a simulation will cre ate and contain all the input parameters of that simulation One extra log file is also created in the working directory but with a different base filename that can be specified by the logfile parameter called nemo log by default see section 3 1 and that will store some runtime information about the simula tions done No replicate or generation counter is added to these files txt these files contain the statistics computed by a simulation and are created CHAPTER 2 THE INIT FILE 20 only when the simulation is asked to see section 4 13 These files don t add any counter string to t
40. adlt mal pi number of males in patch i adlt sexratio adlt sexratio see above off sexratio Table 7 1 Population stat options continued off sexratio offspring sex ratio 7 4 Neutral markers Table 7 2 Neutral markers stat options Stat option Output name Description Note More details about the stats are given in section 5 2 adlt off coa age theta mean within deme coancestry age alpha mean between demes coancestry adlt coa persex adlt thetaFF mean within deme within females coancestry adlt thetaMM mean within deme within males coancestry adlt thetaFM mean within deme between sexes coancestry adlt off adlt off theta as above coa within adlt off adlt off alpha as above coa between lad1t off age theta as above coa matrix age alpha as above age coal deme specific mean coancestry within deme i for all demes Table 7 2 Neutral markers continued on next page CHAPTER 7 OUTPUT STATISTICS 88 Stat option Output name Description age coal j deme specific mean coancestry between demes 7 and 7 for all pairwise comparisons lad1t off age theta as above coa matrix within age coat deme specific mean coancestry within deme 2 for all demes sibcoa prop fsib mean proportion of full sib prop phsib mean proportion of paternal half sib prop mhsi
41. al window or something approaching depending on the program s version gt nemo2 1 0 NEMO2 1 0 22 Jan 2009 Copyright C 2006 2009 Frederic Guillaume This is free software see the source for copying conditions There is NO warranty not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE http nemo2 sourceforge net reading Nemo2 ini setting random seed from input value 213145234 SIMULATION 1 8 POLY_dcost01_ISM CHAPTER 1 INTRODUCTION 9 start 23 01 2009 11 33 27 mode overwrite traits delet fdisp mdisp ntrl LCEs breed_selection 1 store 2 save_stats 3 save_files 4 disperse_evoldisp 5 aging 6 extinction 7 outputs test log delet del fstat dat fstat fsti binary bin data txt replicate 10 10 11 34 32 3000 3000 end 23 01 2009 11 34 40 done CPU time 00 01 11s setting random seed from input value 213145234 SIMULATION 2 8 MONO_dcost01_ISM start 23 01 2009 11 34 40 L This output shows the progress of the simulation with the replicate and generation counters and prints the time when the current replicate started in format hh mm ss and ended and the elapsed computing time hh mm ss This simulation was run on a MacBook 2 4 GHz Intel Core 2 Duo The parameter file used in this example is the one present in the example directory of the distribution package 1 3 1 2 For Windows users You have two options t
42. alues The value at each locus is set by di viding the initial value by two times the number of loci The initial population will then be monomorphic for this trait value unless specified otherwise CHAPTER 5 TRAITS 70 quanti_init_model 0 1 opt If the initialisation model is set to 0 the initial population will be monomor phic for the initial trait values specified previously If set to 1 a random mutational effect is added to each locus on top of the initial value Model 1 is the default quanti_environmental_variance double opt Variance of the environmental deviation of the trait s phenotype Is zero by default no environmental variance A random Gaussian value with mean zero is added to the genotypic value otherwise quanti_output bool genotypes opt If present the phenotypes of the whole population are saved in a text file with one individual per row The genotypic values are added if the environmental variance is not null The data saved is pop P1 G1 age sex home ped isMigrant father mother ID with pop the patch identifier P1 the phenotypic and G1 the genotypic values only if environmental variance is set of the trait age the age class 0 offspring 2 adults sex gender 0 male 1 female home the patch the individual was born in ped the pedigree class check the manual p 53 isMigrant a number telling
43. at_generation 1000 resize_keep_patch 11 5 4 2 3 Note that this reordering will not have any consequence on the evolution of the population unless the migration scheme is different from the island model CHAPTER 4 LIFE CYCLE EVENTS 49 Examples Here is an example of the fusion of two patches into one patch_number 2 patch_capacity 100 resize 1 rank in the life cycle resize_at_generation 1000 resize_patch_number 1 resize_patch_capacity 200 resize_do_fill Using the population parameters only would not lead to the fusion of the two patches as shown in this next example Instead one patch the first one will be destroyed along with the individuals it contains while the carrying capacity of the remaining patch is increased to 200 patch_number 960 2 g1000 1 patch_capacity 960 100 g1000 200 Temporal argument values can be used to model more complex demographic scenario as in this next example patch_number 6 patch_capacity 200 resize_at_generation 100 1000 2000 3000 resize_patch_number 0g0 1 g1000 2 g2000 4 g3000 6 resize_patch_capacity g0 200 g1000 100 g2000 150 g3000 200 resize_do_fill 990 1 g2000 0 Here the population starts with 6 patches of size 200 A massive extinction occurs at generation 100 reducing the population to one patch of size 200 The population then starts growing again from generation 1000 to 3000 with the fission of
44. ation model The same variance is used for all traits unless the full mutation covariance matrix is specified see below quanti mutation correlation double opt The correlation of the effects of pleiotropic mutations when two or more traits are modelled It applies to both the di allelic two traits only and the con tinuous models For the di allelic case the correlation is interpreted as the probability of having the same sign of the mutation effect For the continu ous model the correlation is transformed into a covariance using the value of quanti_mutation variance to build the mutation matrix quanti mutation matrix matrix opt The covariance matrix of the multivariate Normal distribution used to draw the mutation effects in the continuous allelic model Can be used to set dif ferent mutational variances for the different traits This must be a square symmetrical and semi definite positive matrix with trait mutational variance on the diagonal and the mutational covariance off the diagonal This matrix is often referred to as the M matrix quanti recombination rate double matrix opt The recombination parameters are now v2 3 managed by the genetic map See section 5 1 for the details quanti_ init value matrix opt The initial genotypic value of the trait can be set here It is O by default This parameter is valid for the whole metapopulation The LCE quanti_init can be used to set patch specific initial v
45. b mean proportion of maternal half sib prop nsib mean proportion of non sib coa fsib mean coancestry within full sib coa phsib mean coancestry within paternal half sib coa mhsib mean coancestry within maternal half sibs coa nsib mean coancestry within non sib adlt off age ntrl li aj frequency of allele j at locus 2 in the ntrl freq whole population age ntrl 1 Het mean heterozygosity of locus 7 in each patch lad1t off age allnb mean number of alleles per locus in fstat the whole population age allnbp mean number of alleles per locus within demes age fixloc mean number of fixed loci in the whole population age fixlocp mean within demes number of fixed loci age ho observed heterozygosity age hsnei expected demic heterozygosity Nei amp Chesser 1983 age htnei expected total heterozygosity age fis Fig Nei amp Chesser 1983 age fst Fer Ggr Nei amp Chesser 1983 age fit Fyr Nei amp Chesser 1983 lad1t off age fst WH the Weir amp Hill 2002 Fyr estimate weighted fst Table 7 2 Neutral markers continued on next page CHAPTER 7 OUTPUT STATISTICS 89 Stat option adlt off weighted fst matrix Output name age fst WH Description the Weir amp Hill 2002 Fsr estimate age f st1 deme specific Far within deme 7 for all demes age sti j deme specific Fgr between demes i and j for all pairwise comparisons lad1t off age fst WH the Weir amp Hill 2002 Fsr estimate weigh
46. breeding and dispersal in a single step It inherits the param eters of the breed and disperse LCEs For an offspring each parent is randomly taken from the local patch with probability 1 m or from a different patch with probability m where m is the dispersal rate The dispersal rates are thus taken as backward migration or immigration rates in opposition to the forward emigration rates of the disperse LCE This corresponds to the classical Wright Fisher model if the mating system is hermaphroditism mating_system 6 By default exactly K off spring are produced per patch if K is the patch capacity unless the patch is extinct and the parameter breed_disperse_colonizers is specified which limits the number of individuals grown locally from two immigrant gametes The number of offspring produced locally can also be density dependent and set following different growth models using parameters breed_disperse_growth_model and breed_disperse_growth_rate The following features differ from the two base LCE s e backward migration the columns of the dispersal matrix must sum to 1 instead of the rows because Nemo reads the immigration rates column wise element d is the probability to get a migrant gamete from deme i into deme j i being the row number and j the column number e There can be no demographic stochasticity demes always at carrying capacity if the growth model is set to 1 instant growth default value and breed _disperse_colon
47. by the FSTAT program Goudet 1995 as it adds some information about each individuals age sex pedigree and natal patch An example of an output file is given below e GENEPOP genepop Same as for the FSTAT option although it saves the data in GENEPOP format Reymond amp Rousset 1995 ntrl_ save fsti bool opt This tells nemo to save the within patch Far values per locus using the Weir amp Hill 2002 estimates see note below Each line of the output text file contains the values of a specific locus and each column is for a different patch The first line takes the column labels The file extension is fsti ntrl_ save freq allfreq vcomp opt This saves the per patch and per locus allele frequencies default option allfreq or with option vcomp the per locus variance components used to compute the Fsr WC84 i e the a b and c components as described in Weir amp Cockerham 1984 In the first case the file has as many lines as the number of loci and as many columns as the number of alleles per patch denoted avp for allele k CHAPTER 5 TRAITS 65 in patch i In the case of the variance components the file has 4 columns one for each variance component and one for the locus specific Fst Each line also contains the information for one locus at a time The file extension is freq NOTE if the population contains both adult and offspring individuals at the time of writing the file only the offspring ar
48. c e Deleterious mutations stats mutation frequency heterozygosity homozy gosity genetic load heterosis number of lethal equivalents viability by pedi gree classes etc e Dispersal trait stats mean male and female dispersal rates e Population stats patch saturation female and male number per patch sex ratio mean fecundity variance of reproductive output count of migrants effective extinction rate etc The summary statistics are then dumped to a text file at the end of a simulation This file is easily handled by classical statistical packages such as the excellent R for further analysis and graphical representation Alternatively you can save the raw data of the ancestral population in either binary or text file formats The various traits usually provide a way of saving the popula tion genotypes in text files A special binary file format is used to save the whole population information containing all the traits and individuals data and the simu lation parameters Binary files can then be used by Nemo to load a saved population and run a new simulation from it or use it as a source of individuals that have for instance reached a certain level of genetic stability i e burn in population 1 3 Using Nemo Let s assume you have copied the executable file corresponding to your operating system on your disc and that you have launched a terminal window The following guidelines will show you how to launch a simulation o
49. c conditional stochastic if the number of breeding adults is below K 2 use model 7 else use model 3 6 fixed fecundity the number of offspring produced in patch 7 is NG N f f the mean fecundity set by mean_fecundity 7 stochastic fecundity as in 6 but with the total number of offspring drawn from a Poisson distribution of mean equal to N1 breed disperse growth rate decimal opt The patch growth rate used in the logistic growth model 4 19 Breed with selection and backward migra tion name breed_selection_disperse integer age flags adults required and offspring added files NA inherits from breed_disperse selection This LCE aggregates the features of both previous composite LCEs However to perform selection and backward migration with populations of constant sizes there must be some adjustments in the way selection is performed in the case where the mean fitness is too low to allow the patches to be filled with surviving offspring The basic idea is therefore to define a minimum fitness threshold for the individuals If the mean fitness of the adult breeders generation is below that threshold before mating the offspring fitness is rescaled so that the mean patch fitness matches that threshold In other word the threshold is the minimum survival probability offspring in a patch can reach and the scaling factor is ES As soon as the mean patch fitness is above that threshold the scaling factor is reset to
50. ch patch the offspring individuals are randomly chosen to fill the adult containers until the patch carrying capacity is reached Note since the behaviour of this LCE has changed in version 2 0 7 be careful about its position in the life cycle If placed before disperse no offspring will be able to migrate in the population as they already aged The regulation event is not useful anymore after aging but is still proposed in a slightly different flavour see below 4 2 Breeding name breed integer age flags adults required and offspring added CHAPTER 4 LIFE CYCLE EVENTS 33 files NA derived components breed_wolbachia breed_disperse breed_selection breed_selection_disperse Performs mating and breeding of the new offspring generation following the mating system chosen Adults are not removed here see aging above The number of offspring per female depends on the mean fecundity set by mean_fecundity below and may be a fixed number or a number drawn from different random distributions The default distribution is Poisson mating system 1 to 6 Six mating systems are implemented in Nemo The options are 1 promiscuity random mating One male is randomly chosen for each new offspring a female does 2 polygyny One male only mates with all females in the patch This can be changed by setting mating proportion to a value lt 1 in which case one male will monopolise a proportion equal to mating_proportion of the matin
51. chapter 7 for a description of the different output files declared by this LCE Note that no results will be saved if none of save_stats or save files are present in the life cycle stat string CHAPTER 4 LIFE CYCLE EVENTS 52 The string passed to this parameter must contain the stat options defined by the various simulation components A list of these options is given in chapter 7 Note This is the only non sequential parameter the list of arguments is considered as one complete character string stat_log time integer This is the generation recording time of the summary statistics defined by the previous parameter stat_dir string opt This optional parameter is used to specify a path to a directory where to save the stat files It shall not end by a slash character stat_output_compact bool opt Changes the format of the output stat files by suppressing the pretty printing of each column with lots of space between them Instead each value is separated by a single space character The value separator can be changed to a comma with the next option below Use this to save space on disk stat_output_CSV bool opt Changes the column separator from a white space to a comma Implies compact output format stat_output_width integer opt Sets the column with in the output stat files Is 12 characters by default stat_output_precision integer opt Sets the decimal precision in the output stat files Is 6
52. cific LCE e g sequence data before and after disperse This will probably change in future releases Some simulation components automatically upload their different file handlers to the file manager For instance the save stat LCE defines two types of automatic output files one ending with the txt and the other with the _bygen txt extensions see above and chapter 7 to save the statistics recorded during the simulation Other component let the user chose what and when data must be saved on disk see the trait components for e g 4 15 Store Data in Binary Files name store integer age flags unchanged files bin tar bz2 This LCE provides a way to dump all the traits and individual s data to a binary file That file can then be used to initiate a new simulation using the source_pop option in the population parameters Binary files contain all the genetic and individual data plus the whole set of parameters that allowed to generate these data More than one generation of one replicate can be saved in one binary file but there always is one file per replicate By default binary files are compressed with bzip2 by default and put in a tar archive This behaviour can be changed with the parameters described below CHAPTER 4 LIFE CYCLE EVENTS 54 store_dir string opt Used to specify the directory where to save the binary files store generation integer The generation to save in the binary files The last
53. ction in CHAPTER 4 LIFE CYCLE EVENTS 59 one but doesn t add any new parameters Other composite LCEs may also add new parameters see below Because the init file cannot have more than one copy of a parameter the composite LCE and its base LCEs cannot have different parameters values they share the exact same parameters That behavior will change in future releases breed_selection breed_disperse breed_selection_disperse 4 17 Breed with selection name breed_selection integer age flags adults required and offspring added files NA inherits from breed selection This composite LCE performs breeding and viability selection on the offspring gen eration It inherits the parameters from the breed and the viability_selection LCE s parameters as described before No additional parameters are required The following features differ from the base LCE s e Fitness is always absolute e The realised fecundity of a female or male is set accordingly to the sur vival of their offspring allowing the correct computation of the values of the heterosis load and females males realised fecundities and fecundity vari ances e This LCE may be faster than having breed followed by viability_selection in the life cycle when more than one trait are simulated because mutation and recombination are performed on the selected trait before checking for survival Therefore mutation and recombination of the traits not under sel
54. d mysourcepop_001 bin is copied by each replicate of that simulation Now if we want to change that behavior and use a different source population for each replicate we must specify the following set of parameters CHAPTER 3 SIMULATION COMPONENTS 28 replicates 10 source_pop binarydir mysourcepop source_preserve source_replicates 10 source_replicate_digit 3 In that second example each replicate loads a different population mysourcepop_001 bin for replicate 1 mysourcepop_002 bin for replicate 2 etc If the simulation to run has a hundred replicates and we keep the same set of parameters for the source the source population will be changed every four replicates only starting from replicate 25 Replicates 1 to 4 will use data from the population in mysourcepop_025 bin replicates 5 to 9 will use mysourcepop_026 bin and so on until file mysourcepop_049 bin replicates 100 source_pop binarydir mysourcepop source_preserve source_replicates 25 source_replicate_digit 3 source_start_at_replicate 25 Finally loading a population from a trait file is also possible This can be done from a single or different files depending on the type of data The simulation parameters should match the data structure in the source file for optimality The following example loads neutral markers data e g from a field study from a single FSTAT file see section 5 2 for more details and use it to compute the F statistics a
55. duction breed disperse dispersing sex female male opt Specifies the sex of the dispersing gamete used when only females monoe cious individuals are present in demes as for hermaphroditic or self fertilising mating systems models 6 and 4 respectively Should be set to male to model pollen dispersal i e male gamete dispersal to indicate which dis persal matrix must be used to select the right father which in this case is another female hermaphrodite individual possibly in another patch If hermaphrodites are sessile individuals plants and the ovules do not disperse then the breed_disperse_matrix_fem must be set to the identity matrix complete philopatry breed disperse growth model 1 7 opt 1 instant growth patches are filled to their carrying capacity within one generation This is the default model 2 logistic growth the number of offspring produced in patch 7 is given by the classical logistic growth model with Ny Ng rNpg x K Ng K with r the growth rate given by breed_disperse_growth_rate Ng the number of breeding individuals and Ny the numbers of juveniles produced in patch 2 CHAPTER 4 LIFE CYCLE EVENTS 98 3 logistic stochastic the number of offspring is drawn from a Poisson distribution with mean set by the logistic model as above 4 logistic conditional if the number of breeding adults is below K 2 use model 6 else use model 2 5 logisti
56. e The individuals are considered hermaphrodites here that is only the females are used watch the patch size parameters CHAPTER 4 LIFE CYCLE EVENTS 34 mating_proportion decimal opt This parameter is used to set the proportion of random mating in the polygyny and monogamy mating systems and the selfing rate for the selfing case See the mating systems description above for more details The actual proportion of random mating will be 1 mating_proportion on average This can be used to set the degree of extra pair mating when monogamy is modelled for instance mean fecundity integer Mean of the distribution used to set the females fecundity It is used whatever the mating system selected fecundity distribution fixed poisson normal opt The distribution used to set the females fecundity Is Poisson by default The fixed option sets the fecundity of each female equal to the mean see mean fecundity above fecundity _dist_stdev decimal opt Standard deviation used in case the fecundity distribution is set to normal mating males integer opt This parameter sets the number of males that will be available for mating within each patch under polygyny only The value given in argument should be equal to or smaller than the male s carrying capacity Setting it to the carrying capacity is equivalent to setting the mating system to monogamy sex_ratio_mode fixed random opt By default t
57. e class The age class offspring or adults or both that is used when loading a population depends on the one available in the source file and the one that is required by the life cycle events of the current simulation The class to load is determined by finding the first event in the current life cycle that requires a specific age class see chapter 4 on life cycle events Nemo then tries to load that class from the source file Independently of the loading mode if that required age class is not available in the source population the alternate one is used instead i e offspring for adults and vice versa A warning message is displayed if that case happens Using compressed binary files Finally when loading populations from binary files Nemo will automatically check whether the binary fill is compressed If so Nemo will decompress the file read it and recompress it Files saved in an archive will however not be extracted This feature is only possible if one of the two default compress formats is used gz and bz2 and the corresponding programs are available on the system Parameters description CHAPTER 3 SIMULATION COMPONENTS 26 source_pop string opt The path to the population file is given by this parameter The path filename of the source population may contain the special expansion character and format string see chapter 2 to match the sequential parameter arguments values defined in the current configuration file The
58. e section 2 4 below Let s illustrate this by first running Nemo with more than one argument gt nemo2 0 siml ini sim2 txt sim3 Here we have three init files called sim1 ini sim2 txt and sim3 they are all text files the extensions do not matter much here Their parameters are the same as in the previous example This command will produce the following output NEMO 2 0 0 25 Apr 2006 reading sim1 ini reading sim2 txt reading sim3 SIMULATION 1 3 replicate 10 10 10 04 54 100 100 done CPU time 00 01 26s SIMULATION 2 3 replicate 10 10 10 06 36 100 100 done CPU time 00 01 26s CHAPTER 1 INTRODUCTION 11 SIMULATION 3 3 replicate 10 10 10 08 13 100 100 done CPU time 00 01 26s Sequential parameters As an example of sequential parameters let s assume the first file sim1 ini has the following parameter with several arguments patch_capacity 5 10 20 This will add two more simulations to the three previous ones gt nemo2 0 sim1 ini sim2 txt sim3 erent reading sim1 ini reading sim2 txt reading sim3 SIMULATION 1 5 replicate 1 10 10 19 22 88 100 gt Pop extinction replicate 3 10 10 19 25 74 100 gt Pop extinction replicate 4 10 10 19 26 84 100 gt Pop extinction replicate 7 10
59. e used ntrl_output_dir string opt This parameter specifies a specific path used to save the genotype and fsti output files Should not end with a slash ntrl_output_logtime integer opt This is the generation periodicity of the output files or the generations at which the files should be saved if provided as multiple values in an array Note about reading an FSTAT file as discussed in section 3 2 1 it is possible to load a population from genetic data saved in an FSTAT file That file can use the original or the extended file format as described here The original file format does not include the age sex ped and origin loci Here is an example of a neutral genotype output file the file format is inherited from the FSTAT file format Goudet 1995 5 9 20 2 loci loc2 loc3 loc4 loc5 age sex ped origin 1414 1019 2002 0820 0307 0814 0219 2002 2020 0307 0808 0217 1902 0820 0907 0820 0209 1902 0805 0918 A 0307 1308 0220 0401 0115 0905 1213 0302 0312 0506 4 eno 2017 1010 2013 1812 1505 401 5 2017 1008 2013 1811 1505 4 1 2 3 e Aerer Orre PR GO KM KX KA W KK KX KA ae ES Qn 0 RS a rr R R FA Re N N The first line contains the population number 5 pops here the number of locus 5 4 which corresponds to the number of columns saved minus the first one the CHAPTER 5 TRAITS 66 maximum number of alleles per locus 20 and the number of digits used to write
60. e will produce the following files with representing the replicate number from 01 to 10 logfile log test example log test stat example_bygen txt test stat example txt test ntrl example_ dat test delet example_ del test binary example_ bin bz2 More elaborate examples can be found in the example folder of the installation package Chapter 7 Output Statistics The summary statistics computed during the course of a simulation depends on the options given to the stat parameter of the save stats LCE see section 4 13 The options available are declared by the various simulation components the traits and the life cycle events The complete list of these options are given below for each component A typical stat option string as found in the init file builds like this stat fstat off delet viability disp demography which will result in the computation of the F statistics for the offspring and adults the statistics for deleterious mutations on the offspring age class the mean viabilities the mean dispersal rates and additional statistics describing the population state All these options are described below in section 7 2 Note that if one of the component stat option is present in the stat parameter argument but the component itself is missing this will end the initialisation process of the simulation and abort the program An example is given here assuming the dispersal trait is missing but the dis
61. ea awe a ee a a Ea eG ae ill 63 67 71 74 76 TT 78 78 78 79 79 80 Chapter 1 Introduction 1 1 Overview Nemo is a forward time individual based genetically explicit and stochastic simu lation program designed to study the evolution of life history phenotypic traits and population genetics in a flexible meta population framework Nemo implements a recombination map on which loci coding for different types of traits can be placed together The evolving traits provided are sex specific dispersal rates universally deleterious mutations quantitative traits Dobzhansky Muller incompatibilities and neutral markers e g SNP microsatellites It also allows for the simulation of the dynamics of an endosymbiotic parasite vertically transmitted causing cytoplasmic incompatibility Wolbachia The number of populations individuals per population or loci per trait to simulate are only restricted by hardware capacities Nemo is highly optimized to run in batch mode and a parallel computing version is part of the release thus making it a very flexible and powerful simulation tool Nemo s framework is coded in C and has been designed to be easily extended and include new evolving traits or population features Availability Nemo comes free of charges and is distributed under the GNU Gen eral Public License GPL2 Binaries and source code are provided for the Linux MacOSX and Windows platforms Nemo is coded in C and runs on
62. ection are performed on the surviving offspring only breed_selection fecundity fitness bool If this parameter is set present in the init file the selection mode is changed from acting on offspring survival to act on the number of offspring produced by each female In other words with this mode it is the fitness of the female that matters rather than that of the offspring The mean value of the fecundity distribution is multiplied by each female s fitness when drawing its number of offspring produced This works best when the mean fecundity is large because CHAPTER 4 LIFE CYCLE EVENTS 56 only integer numbers of offspring can be produced which is problematic when the mean of the Poisson distribution is too low e g a fitness of 0 25 and a mean fecundity of lt 4 will cause many more females to have no offspring than if the mean fecundity is 10 By having a too low mean fecundity one looses precision in the selective process and selection will be stronger 4 18 Breed disperse gametic migration name breed_disperse integer age flags adults required and offspring added files NA inherits from breed disperse Note since version 2 3 the dispersal parameters that are inherited from the disperse LCE must now be pre pended with breed_disperse instead of dispersal as in the original LCE For instance dispersal_rate becomes breed_disperse_rate disper sal_matrix becomes breed_disperse_matrix etc This LCE performs
63. ectory random seed integer opt The seed of the random generator can be specified with this parameter The upper value is system dependent but should not be more than 4 294 967 295 on a Mac By default the random seed is set by the clock time of the computer i e number of seconds since an arbitrary date in the past usually around the 1970 s postexec_script string opt This parameter is used to specify the path to a shell script that will be executed once all the simulations have been processed The script will be executed using a system call with the following command sh my_script sh postexec_args string opt This parameter is used to add an argument to the above script when executing it Be aware that the expansion character will not be expanded if present in the argument string and should thus be avoided 3 2 Population name population files NA stats pop demography migrants kinship and more see chapter 7 patch_number integer opt Number of patches in the population CHAPTER 3 SIMULATION COMPONENTS 24 patch_capacity integer matrix opt Carrying capacity of each patch K this is the number of males and females If given as a unique value all the patches have the same size with equal numbers of males and females May also be given as a matrix parameter containing the vector of the patches size In that case the length of the vector will give the number of patches in the
64. ed fsib mean proportion of offspring born from an inbred mating between full sib brother sister individuals ped self mean proportion of offspring born from the mating of selfed parents migrants emigrants mean number of emigrants per patch immigrants mean number of immigrants per patch residents mean number of residents per patch immigrate effective immigration rate computed immigrants immigrants residents colonisers mean number of immigrants per extinct patch colonrate effective colonisation rate of extinct patches migrants patch emigr p number of emigrants from patch i resid pz number of residents in patch 2 imrate pi effective immigration rate into patch 1 computed as anal colo pz number of colonizers of patch 7 is 1 if patch wasn t extinct A value of 0 means the patch was extinct but not recolonized Table 7 1 continued on next page CHAPTER 7 OUTPUT STATISTICS 87 Stat option Output name Description pop same as demography off adlt sexratio and extrate together pop patch off adlt fem pi number of females in patch i off adlt mal pz number of males in patch i age pat Chi time since last extinction of patch i patch avrg age mean time generation since last extinction of a patch extrate proportion of extinct patches in the population off adlt fem patch off adlt fem pi number of females in patch 2 off adlt mal patch off
65. educed internally CHAPTER 4 LIFE CYCLE EVENTS 36 dispersal_model 1 2 3 4 opt The dispersal models implemented so far are 1 Migrant pool Island model If the migration rate is m the probability to disperse to any n 1 non natal patch is Ta while the probability to stay at home is 1 m Propagule pool Island model In that modified version of the Island Model each offspring in a patch has a probability my to move to the same assigned patch With probability Peg they will move to any patch but their home or propagule assigned patches With probability 1 m they will stay home The propagule patches are reassigned every generation Stepping Stone model This is the one dimension Stepping Stone model By default the patches are placed on a circle ring population and the dispersers can only move to one of the two adjacent patches This model can be changed by using different border models see below Lattice model Patches are placed on a squared grid or lattice and dispersers can move to at least four adjacent patches set by the disper sal_lattice_range parameter below This option must be followed by the dispersal_lattice_model and dispersal_lattice_range parameters The number of patches in the population must be a square number The dispersal_model parameter may be omitted when providing the dispersal matrix or reduced matrix dispersal lattice range 1 2 opt Sets the number of
66. exact rate of change of the local phenotypic optima will thus be set depending on the amount of phenotypic variation in the population To set the actual rates the two next parameters are necessary to measure the phenotypic standard deviation selection_std_rate_set_at_generation integer opt This is the generation at which the phenotypic standard deviation must be measured to set the relative rate of change of the phenotypic local optima CHAPTER 4 LIFE CYCLE EVENTS 43 selection_std_rate_reference patch integer opt The phenotypic standard deviation of the traits under shifting environmental conditions can either be the average over all patches or set from a single reference patch This parameter is used to specify that reference patch The population average of patch specific phenotypic standard deviations will be used if the parameter is not present in the init file not set 4 8 Extinction and Harvesting name extinction integer age flags unchanged files NA This LCE is used to either cause the random extinction of patches in the population following the extinction rate or reduce their size by a given amount or proportion i e harvesting If a patch goes extinct it is completely emptied of all the individ uals present This LCE only acts on the content of the patches it never modifies their capacities see resize for that An extinction threshold can also be set as a percentage of the patch capacity and is used to c
67. f the mutational effect s logarithm These two parameters are specified by delet_effects_dist_param1 and delet_effects_dist_param2 re spectively Note that the distribution is truncated to the right no value greater than 1 is allowed delet_effects_mean decimal Mean effect of the deleterious mutations Also known as the selection coeffi cient of the mutations Is used to parameterize the effect sizes distribution delet_effects_dist_param1 decimal opt Extra parameter used for the description of the distribution of mutational effects This is the shape of the gamma distribution or the logarithmic mean effect in case of the log normal distribution delet_effects_dist_param2 decimal opt Second extra parameter used for the description of the distribution of muta tional effects This is the scale of the gamma distribution or the logarithmic standard deviation in case of the log normal distribution delet_dominance_mean decimal Dominance coefficient alternatively the mean of the distribution of dominance coefficients of the deleterious mutations CHAPTER 5 TRAITS 73 delet_dom_coef decimal Equivalent to delet_dominance_mean kept for backward compatibility delet_sel_coef decimal Equivalent to delet_effects_mean kept for backward compatibility delet_continuous_effects bool opt Deprecated since version 2 0 7 delet_fitness_model 1 2 Sets the fitness model used to compute the individual viability
68. from the dele terious genome the trait phenotype 1 Multiplicative model The individual fitness or viability is computed as the product of the fitness of each locus W 1 1 s 1 hasina where n is the number of homozygote loci and na the number of heterozygote loci s is the selection coefficient and h the dominance coefficient 2 Additive model Here mutations act non independently on fitness this may be viewed as an epistatic model The individual fitness is W 1 n s n2hs Symbols has same meaning as previously W is truncated at 0 fitness can never be negative here delet fitness scaling factor integer opt This parameter s value is used as a scaling factor for the individual s pheno type i e its viability is multiplied by this value delet_save genotype bool opt Parameter used to save the population genotypes in a text file with the del extension The first line holds the column labels Each line starts with the population identifier followed by one column per locus plus the age sex pedi gree class and patch of origin of each individual The allelic values are 0 for the wild type allele and 1 for the deleterious allele For the cases where mutational effects are continuously distributed the second row holds the se lection coefficient homozygous effect of each locus and the third one holds the heterozygous effects of each locus delet_genot dir string opt This paramete
69. generation will always be saved whatever the value given here store_recursive bool opt This option will tell the program to use the store_generation value as a generation logging time The binary files will thus contain several generations store_noarchive bool opt This option suppresses the archiving of the binary files store nocompress bool opt This option will suppress the compression of the binary files store_compress_cmde string opt The program used to compress the binary files is by default bzip2 You can change this default behavior by specifying a alternative program or path to that program to use here store compress extension string opt The alternative used with the previous parameter will probability use a differ ent file extension than bz2 Use this parameter to specify that alternative extension store archive cmde string opt Similarly to the compression process an alternative archiver program can be specified here to avoid the use of tar store_archive_extension string opt The file extension used by the alternative archive program can be specified here 4 16 Composite LCE Composite life cycle events are LCEs that inherit the properties parameters of other LCEs the base LCEs and extend or sometimes redefine their function alities For instance breed_selection inherits the parameters of the breed and viability_selection LCEs and performs both breeding and viability sele
70. gs within a patch while the remaining matings are shared by all other males The number of mating males may also be changed below with the mating males parameter in which case the mating male for a given female is randomly chosen within the mating_males first males of a patch 3 monogamy Each female mates with one male only and vice versa If the number of males is less than that of females some males will mate with more than one female In the reverse case however if there are more males than females some males will not reproduce at all A given proportion of random mating can be achieved by setting the mating_proportion param eter to a value lt 1 Each female will then have on average a proportion of 1 mating_proportion of its offspring descended from a random male in the population 4 selfing hermaphrodite Only females are used in that case If mat ing_proportion 1 all offspring are produced by self fertilisation other wise a proportion of 1 mating_proportion of the offspring are produced by randomly crossing two females together 5 cloning Equivalent to selfing but without recombination Individuals are produced by first copying the mother s genes and then computing mutations The mating_proportion parameter is used in the same way as under selfing 6 random mating with selfing This corresponds to what is called the Wright Fisher model where individuals may self with probability 1 N N patch siz
71. haploid by default Incompatibil ities come by pair and pairs of locus are contiguous on the chromosome s The recombination rate between each locus is set below dmi is _haploid bool opt Can be used to change the ploidy of the trait Is set to true by default The trait will be diploid is this is set to 0 false dmi_mutation rate decimal Per locus mutation rate Mutations are both way 0 gt 1 amp 1 gt 0 dmi recombination rate dmi genetic map dmi random genetic_map opt Recombination is handled by the genetic map All genetic map parameters apply See section 5 1 CHAPTER 5 TRAITS 75 dmi_genot_table matrix This table sets the fitness of each pair of locus relative to the wild type It must be set for each pair explicitly no repetition of patterns for now The structure is one row per incompatible pair and one column per genotype There is here a slight difference for the haploid and diploid versions In the haploid case the fitness of all 4 genotypes must be given There are two incompatible pairs aB and Ab The fitness associated with each genotype is written in the following order for haploids and one pair AB aB Ab ab for diploids and one pair AABB AABb AAbb AaBB AaBb Aabb aaBB aaBb aabb For the diploids 9 genotypic values must be given We do not distinguish between single locus heterozygotes i e Aa aA The incompatible pair is the middle one
72. he number of available values a parameter can take may be specified from case to case e decimal argument may be a floating point value The following forms are equivalent 0 0001 0001 or 1e 4 e string argument is a character string that may contain white spaces e matrix special argument that is enclosed by inside these brackets each row of the matrix is also enclosed by two brackets see section 2 3 for details and examples 12 CHAPTER 2 THE INIT FILE 13 2 2 Special characters Here is a list of the reserved characters and their meaning during the process of reading and parsing the input parameters file e comment any character that follows the comment character is removed until the end of the line is found If a starting block comment string is found within a commented line it is treated as such see below e block comment any line of text enclosed by those two characters strings is recursively removed from the init file A block comment can also be specified on a single line e line continuation the line that immediately follows that character is appended to the current line and the two lines are treated as one This is particularly useful to split a sequence of argument values over several lines see the matrix example below e matrix row1 any argument value starting and ending by two enclosing curly braces is considered as a matrix argument see next section e
73. he sex of an offspring is randomly set unless the individuals are considered hermaphrodites and thus the offspring sex ratio usually varies from one generation to another The fixed option proposed here sets the sex ratio to exactly 1 1 4 3 Breeding with Wolbachia name breed_wolbachia integer age flags adults required and offspring added files NA inherits from breed This is also a derivative of the first breeding LCE it thus inherits the previous parameters and defines several parameters for the simulation of Wolbachia infections See the Wolbachia trait for more details CHAPTER 4 LIFE CYCLE EVENTS 39 wolbachia fecundity_cost decimal The fecundity of an infected female as specified by parameter mean_fecundity is reduced by an amount of 1 sy sf being the cost to pay when infected by Wolbachia wolbachia incompatibility cost decimal A zygote issued from a infected male gamete and an uninfected female gamete must pay the cost of cytoplasmic incompatibility caused by the parasite This cost is the amount of reduction in the survival probability of the offspring wolbachia inoculum size integer Wolbachia can be inoculated to a specified number of adults specified by this parameter This number represents the number of females and the same num ber of males that will be inoculated in one deme of the population randomly wolbachia_inoculum_time integer Generation at which the population will be
74. heir filenames bin these files contain the complete set of individual data for each replicate of a simulation Their filename thus contain the replicate counter appended after the base filename See section 4 15 for more details about the binary output files and how they are handled freq quanti delet etc each component especially traits define their own output files and extensions making it clearer what data is recorded in which file See the next chapters for details Important Note To make sure the file manager of Nemo notifies the different simulation components at time of saving you must include the save_files life cycle event see section 4 14 in the life cycle otherwise no files will be written for a simulation See chapter 4 to understand how this is done In absence of this life cycle event only one type of file is automatically written during a simulation this is the log simulation file holding the simulation parameters and some info about the simulation value of the seed of the random generation elapsed time and CPU time used Chapter 3 Simulation Components This chapter presents the various simulation components and their parameters It is through these parameters that you can select which components are part of a simulation or not Two components are mandatory the simulation and population components Besides these two it would make sense to select at least a basic sequence of
75. hole metapopulation option relative_global 4 7 1 Multi trait selection The traits under selection must be passed to the selection_trait parameter enclosed within parentheses and coma separated i e traitl trait2 and likewise for the selection models associated with each trait in the same order i e model_traitl model_trait2 To specify a model with selection on the delet and quant traits the following set of parameters would be necessary CHAPTER 4 LIFE CYCLE EVENTS 41 selection_trait delet quant selection_model direct gaussian parameters specific to the Gaussian selection model selection_trait_dimension 1 selection_variance 4 selection_local_optima 5 The fitness value of an individual is then given by the product of the fitness values of each trait 4 7 2 Fixed selection model parameters selection base fitness decimal opt Base fitness of the population Wo selection _lethal_equivalents decimal opt Number of lethal equivalents present in the population A selection_pedigree_F matrix opt The values of F for each of the 5 pedigree classes present in Nemo Must be an array of size 5 The 5 classes are outbred between patches might experience heterosis outbred within patches half sib full sib and selfed individuals 4 7 3 Gaussian and quadratic model parameters selection_matrix matrix opt This is the selection matrix w used to set the strength stabil
76. how many parents of the individual are immigrant from same or different source patch father mother and ID are individual id s unique numbers assigned to individuals that can be used to check for pedigree The columns P2 and G2 are added when two traits are modelled If the option genotypes is passed the allelic values are also saved and 2x number of loci x number of traits columns are added to the file quanti logtime integer The timing at which phenotypes should be saved or the generations at which the files should be saved if provided as multiple values in an array quanti_dir string The file directory relative to the root_dir directory CHAPTER 5 TRAITS 71 5 4 Deleterious mutations name delet files del input output phenotype a real value in 0 1 interpreted as the fitness value of the individual Deleterious mutations are mutations that reduce the fitness of their carrier This translates into a lower survival probability of the offspring bearing more mutations when applying viability selection on them see section 4 7 Deleterious mutations are coded by bi allelic loci with value of 0 for the wild type healthy form and 1 for the deleterious form The strength of the deleterious effect of each mutation i e strength of selection and its dominance can be set using two different models constant over loci or following a given distribution over loci The selection and
77. if the environmental variance is different from zero age qi Qst index of population genetic differenciation for the quantitative trait calculated from Va and Vb as Qst yin ST V 2Va age qij cov average genetic covariance within patch between trait 7 and trait j present only if more than 2 traits are modelled adlt off quanti eigen age q evalz eigenvalues of the D matrix the covariance matrix of population means age q evectij loadings of the th eigenvector of the D matrix lad1t off quanti eigenvalues age q evalz eigenvalues of the D matrix the covariance matrix of population means adlt off quanti eigenvectl age q evectli loadings of the first eigenvector of the D matrix lad1t off age qi pj mean phenotypic value of trait in quanti mean patch patch 7 adlt off age Va qi pj additive genetic variance of trait i in quanti var patch patch j age Vp qi pj phenotypic variance of trait in patch j only if the environmental variance is not zero adlt off quanti covar patch age cov qij pk genetic covariance between trait 7 and j in patch k adlt off quanti eigen patch age qevalz p7 eigenvalues of the G matrix in patch j genetic covariance matrix age qevectij pk loadings of trait j on eigenvector i of the G matrix in patch k adlt off age qevalz p7 quanti eigenvalues patch eigenvalues of the G
78. in Features Nemo is a forward time simulation program This means that the population state is evolved forward in time from generation O to generation T through successive generational iterations of the life cycle The life cycle is also composed of a suc cession of events chosen by the user The individuals in the simulated population are run through this life cycle They will do so only once during their lifetime as the kind of organism modeled so far is semelparous i e reproduce only once and then die like pacific salmons for e g The fate of an individual may depend on its traits value or phenotype For instance during viability selection an individual will survive only if its viability trait e g deleterious mutations gives it a chance to win the viability lottery Nemo allows the updating of parameter values during a simulation by using temporal parameter arguments see section 2 6 The population state i e number and size of the patches or any other model component can be changed through time Patches can also be merged or split see the resize life cycle event to model population fusions fissions Nemo offers many different kind of life cycle events see below that allow the user to set up many different population evolution models or simply interact with the simulation data For instance Nemo can load simulation data in various format see subsection 3 2 1 to start a new simulation or just perform some extra genetic ana
79. its unique patch into two smaller ones first i e do_fill is true Two empty patches are added at generations 2000 and 3000 do_fill 0 while the patch capacity increases from 100 to 200 over 2000 generations bringing the population to its original state Note that the temporal specifiers all start with 0 as expected by default which sets the argument values for the first time resize will run that is at generation 100 in the above example The next temporal values must be set at times corresponding to those within the resize_at_generation array argument That parameter can also be a temporal argument however the array form is preferred for its compactness The following example illustrate this point both statements are equivalent resize_at_generation 100 1000 2000 3000 resize at generation g0 100 g1000 1000 g2000 2000 g3000 3000 CHAPTER 4 LIFE CYCLE EVENTS 50 4 11 Cross Design NCI name cross integer age flags adults required offspring add files NA The cross LCE lets you perform a North Carolina I crossing design or half sib full sib design of the population at a given time point during a simulation The LCE creates sire x dam x offspring offspring in each patch of the population It is thus advised not to set the numbers of sires or dams higher than the number of males or females present in the patches This will also replace any offspring previously present in the patches a warning is issued Sire
80. iz_chromosome_num_locus array opt The number of loci per chromosome can be varied using this option giv ing locus numbers in an array The sum of the array must then be equal to the total number of loci of the trait The array must have as many ele ments as the number of chromosomes specified by one of the map options pre fix_random_genetic_map or prefix_recombination_rate This option is not used when fixed maps are specified with prefix_genetic_map see note above prefix genetic map resolution decimal opt The map resolution is by default the centimorgan cM The map positions specified by prefix_genetic_map or prefix_random_genetic_map thus refer to that scale The scale can be changed here by specifying the corresponding reduction of scale Thus prefix_genetic_map_resolution must be smaller than 1 and for instance a value of 0 1 means the resolution is changed to the mili Morgan i e a distance of 1 then corresponds to a recombination rate of 0 1 instead of 1 between two loci The interpretation of the distances between loci thus depends on this scale The map resolution applies to all chromosomes and all traits equally If a trait changes the map resolution all trait s maps are rescaled to the smallest scale CHAPTER 5 TRAITS 63 5 2 Neutral markers name ntrl files dat input output phenotype none Neutral markers are genetic markers such as microsatellites or SNPs which are not affected by selection
81. izers is unset e Deme extinctions may cause the program to hang indefinitely if im migration into an extinct deme is impossible e g because of source patch extinction or zero immigration set in the dispersal matrix CHAPTER 4 LIFE CYCLE EVENTS 57 e An extinct deme will be instantly recolonised in a single generation unless the number of immigrants is capped with breed_disperse_colonizers or a growth model is specified e Two dispersal matrices can be used for hermaphrodites to model pollen migration i e fecundation of local ovules with immigrant pollen without ovule migration see breed _disperse_dispersing sex e Mating systems 2 polygyny and 3 monogamy can not be used here e This LCE can be used to mimic the Wright Fisher model when the mating system is set to 6 random mating with selfing rate x e This LCE is much faster than having breed followed by disperse in life cycle be cause exactly N offspring are produced and not x f f being the females mean fecundity Usually f should be greater than 2 to avoid too much demographic stochasticity especially with small patch sizes breed_disperse_colonizers integer opt This parameter is used to restrict or set the number of individuals that will re colonise an empty patch to a different value than the carrying capacity of that patch That number is sex specific the actual number of colonisers will be twice the value for dioecious individuals biparental repro
82. izing selection on a set of quantitative traits within a patch The w matrix is a square symmetrical positive semi definite covariance matrix The diagonal elements set the strength of selection on each trait selection variance while the off diagonal elements set the strength of correlated selection on pairs of traits selection covariance These values will be applied to all patches equally as only one selection matrix can be specified per simulation selection_variance decimal matrix opt This sets the variance or diagonal elements of the selection matrix w A single value will be interpreted as an identical selection parameter for all traits in all patches A matrix argument can also be passed to change the selection variance among demes and traits This matrix has at most as many rows as CHAPTER 4 LIFE CYCLE EVENTS 42 the number of patches in the population and as many columns as the number of traits modeled When a smaller number of patch values are provided the values will be recycled to fill the patch specific selection matrices Similarly for the trait values although here only a single value is accepted will copy the value to all traits selection_correlation decimal matrix opt This specifies the correlated effect of selection on the different traits This is NOT the same value as you would use in the selection matrix i e covariances A matrix argument can also be provided to set the patch and trait specific val
83. lag values associated with each LCE see Table 4 1 These age flags tell which individual container will actually contain individuals after having executed the corresponding LCE during the life cycle and which age class is needed by an LCE This will help you design a proper life cycle CHAPTER 4 LIFE CYCLE EVENTS 32 Table 4 1 Modification of the population age state caused by the LCEs in the basic life cycle means that age class is added to the population by the LCE while means the LCE will remove all individuals of that age from the population x means the LCE will modify the state of that age class required means that age class is the required age class for the LCE and will be loaded first whenever that LCE begins the life cycle LCE Offspring Adults aging move to adults breed required Cross required disperse x required extinction x x regulation x x resize x x selection x required 4 1 Aging name aging integer age flags removes the offspring flag files NA aging moves all individuals from their age class to the next and performs patch regulation at the same time For now only two age classes are present the offspring and the adults Therefore aging moves the offspring to the adults age class and all the adults are removed they die No other LCE removes the adults from the population It is thus very important to add this LCE to the life cycle For ea
84. ld produce exactly the same result given the sequence of LCEs is conserved see the following examples The only change is the population state at the beginning and the end of the cycle aging 1 disperse 1 breed 2 aging 2 disperse 3 breed 3 Writing the life cycle as above does not ensure that these LCEs will all be loaded into the life cycle as some of them define additional mandatory parameters that must be present in the init file as well The breed and disperse LCEs define such 78 CHAPTER 6 EXAMPLES 79 mandatory parameters The following example will allow to completely build the life cycle breed 1 disperse 2 aging 3 mating system 3 monogamy mean_fecundity 3 mating_proportion 0 8 20 of extra pair matings dispersal_model 2 Island Model with propagule pool migration dispersal_propagule_prob 0 3 30 of propagule dispersers dispersal_rate 0 125 6 1 2 Adding outputs The previous basic life cycle misses two important features It does not record statistics and does either not write any output files To do so you have to add the following LCEs save_stats and save_files breed 1 save_stats 2 save_files 3 disperse 4 aging 5 This way both the adults and offspring statistics are computed and the various files declared by the simulation components are saved to disc Which age classes are present in the population at the time of statistics recording and file writing will de
85. lternative set of argument value identifiers as a character string The format string is enclosed with two single quotes and is composed first of an optional dot followed by a mandatory integer number and finally followed by an optional character string enclosed with two square brackets The optional dot and character strings are mutually exclusive Here is an example of each possible option 4 3 or 2 AaAbAcBaBbBc The format string is placed in between the expansion character and the sequential parameter number like this TAT 3 2 or 2 AaAbAcBaBbBc 3 supposing we have three sequential parameters in an input file The mandatory integer value of the format string is the width of the argument name string For instance 4 4 1 means that the values of sequential parameter no 1 will be written on 4 characters with leading zeros A value of 10 for that parameter will thus be added to the filename string as 0010 The dot preceding the width specifier simply indicates that only the decimal part of the argument value must be taken with trailing zeros In the example above a value 0 1 for sequential parameter no 2 will be added as 100 to the filename string Finally a set of character strings can be specified as in the last example above These characters will be used sequentially as replacement values for the actual parameter values found in the input file The width specifier tells h
86. lysis Nemo can also use genetic markers data to seed a simulation It is thus possible to run simulations based on real field experimental data The number of traits an individual can carry is also up to the user Individuals without any trait can be used to simulate simple demographic models The number of Life Cycle Events LCE composing the life cycle are only limited by their availability These simulation components are added following the needs of the users developers of the Nemo framework and we hope their number will increase with future versions So far the currently available components are as listed below e Life Cycle Events LCE breeding reproduction of dioecious or monoecious individuals viability selection trait and environment dependent fitness values dispersal forward and backward migration combinations of those LCEs see section 4 16 ageing non overlapping generation population regulation ceiling model population growth logistic exponential etc section 4 18 population extinction and harvesting can be patch specific CHAPTER 1 INTRODUCTION 4 population modification fission fusion addition of patches crossing design with half sif full sib design NCI and more e Mating systems the breeding LCE allows for the following mating systems random mating promiscuity polygyny number of mating males can vary monogamy hermaphroditism
87. n e e ecra e paee ie a A AR 4 11 Cross Design NCI on kk ec hh Ghd be A EES Ala Populaticoa Reculdtiom 24 66 eae RE Oe Cw eS De Ole Dare Sial cc bw ce eh PSEA ed Ree eRe OH ES Ali Baring Files s ec Ble oe ee ee Seed heeds Be eda eed 4 15 store Data in Binary Piles oo corso Ree ea CL Composte LCE AA 4 17 Breed with selection oo e 6242 ba eee ees 4 18 Breed disperse gametic migration 4 19 Breed with selection and backward migration 5 TRAITS Gl The Genetic map ee ee eee dor a ee il 23 25 30 32 32 34 39 38 38 39 40 41 41 43 44 44 45 45 47 50 51l 51 53 53 54 59 56 58 60 CONTENTS Ga in III GS Quantitative TW iaa RE AE od 54A Deleferions mutations c oo cis orar er des 5 5 Dobzhansky Muller Incompatibility loci 2 0 0002 5 6 Dispersal genes 224448 ea ea 24 082 208 2 240 ut EPICA sees es a a ada R ee 6 EXAMPLES UL EMS o a A a A a Bee 6 1 1 A basic hS eyele o sirara pinarars rcgio rodi les Adding DUES cs saa pes atad E a A AAN EE E ee ee 6 3 A complete example 1 7 OUTPUT STATISTICS E RR EE MI ee Saeed ee eed wees To SOOO ea et eh ae ae bees HAS ORS See RE p EX T A Neutral mark rts o sc ee eR Re REE ee we Ee we RE fay Quantitativ traits s e eR Da a ee a 7 6 Deleterious mutations e mn e 7 7 Dobzhansky Muller Incompatibilities DMI E AA A AI TIO Wolbachia c e r ro So
88. n model It ranges from the classical island model with evenly distributed patch sizes and dispersal rates to a spatially explicit population model with different sex specific and patch specific sizes and dis persal rates This flexibility is achieved thanks to the matrix parameters that the user can pass to Nemo and which allow to design any kind of population model see chapter 3 for more details Extrinsic population extinction rate can also be added to model extinction recolonization dynamics as well as stochastic variation of pop ulation sizes i e harvesting Furthermore as the model is fully stochastic patch sizes may vary during a simulation as a result of pure demographic stochasticity up to the point of population extinction Here the mean female fecundity is key to set the level of population saturation and demographic stochasticity The population regulation mechanism uses a ceiling model when migration is forward as in the dis perse LCE That is the total number of individuals present in a population at time of regulation is reduced to its carrying capacity for each sex Specific growth rates can be used when backward migration is modeled with the breed_disperse Popula tion bottlenecks and other variation of the population model may also be modeled with temporal parameter values or the use of the resize LCE In summary Nemo allows for the following population features e patch specific and sex specific population sizes patch
89. n of new components such as new evolving traits with their specific genetic architecture and new life cycle events while taking advantage of the simulation management features offered by the framework i e input output management interaction with existing components etc The basic coding procedures are described on the coding documentation web site http nemo2 sourceforge net Acknowledgments The parallel computing version Nemo_MPI has been de veloped in collaboration with Dr Jacques Rougemont at the Swiss Institute of Bioinformatics using the Message Passing Interface MPI standard http www mpi forum org allowing to run simulations on cluster environments such as the Vital IT cluster at SIB http www vital it ch That parallel version uses the Scalable Parallel Random Number Generators library SPRNG http sprng cs fsu edu as a source of random numbers The regular Nemo version implements a random number generator i e the Mersene Twister provided by the GNU Sci entific Library GSL http www gnu org software gs1 as well as several other mathematical routines defined in that library Nemo was initially developed as part of the main author s PhD work at the Department of Ecology and Evolution at the University of Lausanne http www unil ch dee Alistair Blachford provided a first version of the bitwise recombination algorithm Sam Yeaman helped with proofreading and debugging CHAPTER 1 INTRODUCTION 3 1 2 Ma
90. n your desktop computer on both nix flavored operating systems and Windows Guidelines to launch a parallel job on a computer grid or cluster environment are not provided here These will vary according to the type of infrastructure you have access to CHAPTER 1 INTRODUCTION 8 1 3 1 Launching Nemo from the command line 1 3 1 1 For Linux and Mac OS X users On Mac OS X the terminal application called Terminal app is located in the Applications Utilities directory on your hard drive Simply double click to launch it Then whatever your operating system is we assume you have installed the executable file nemo2 x y in a folder somewhere on your file system and that you set your working directory to that place using the cd command The following commands will allow you to run a simulation First lets have a glance at what is in the directory using the 1s command gt ls nemo2 x y Nemo2 ini So we have the executable file nemo2 x y and a configuration file Nemo2 ini Now if we type the following command Nemo will automatically search for the Nemo2 ini file in the local directory and try to initiate a simulation from it gt nemo2 x y The characters in front of the executable filename simply means the program file is to be searched in the local directory rather than in one of the directories specified by the PATH environment variable This command will produce the fol lowing output to your termin
91. neighbouring patches used for dispersal in the lattice dis persal model The dispersal probabilities to these adjacent cells are m 4 in the first case and m 8 in the second 1 2 4 adjacent patches up down left and right 8 adjacent patches as 1 plus the diagonals dispersal border_model 1 2 3 opt In the stepping stone and lattice models i e 1D and 2D lattices three dif ferent ways of dealing with the world edges exist 1 2 Torus This is the doughnut world edges are connected together It has thus no boundaries eliminating any edge effects Reflective boundaries The borders of the lattice 1D or 2D are reflec tive Dispersers from the border cells cannot move beyond the border Border cells have thus less cells connected to them and their dispersal CHAPTER 4 LIFE CYCLE EVENTS 37 probabilities to the adjacent cells are higher e g m m 3 or m 5 de pending on the dimension and range of the lattice No dispersers are lost outside the lattice 3 Absorbing boundaries Dispersers from the border cells of the lattice are lost if they choose to move beyond the border The dispersal proba bilities of a border cell are not modified dispersal_propagule_prob decimal opt Sets the probability that a disperser will move to the propagule assigned patch in the dispersal model 2 dispersal_matrix matrix opt This matrix parameter is used to specify the dispersal matrix of the model It must be pa
92. ng from the first until they reach their carrying capacity If do_flush is not set and the backed up individuals are not in sufficient number the filling procedure will stop before all patches are filled which will happen if the total population size is increased If not set new patches will be empty and undersaturated patches will remain as such and be filled by breeding and immigration in subsequent generations resize do regulate bool opt If set the patches will be regulated to their carrying capacities This will affect the offspring and adults similarly The patch sizes will be at most equal to their carrying capacities Regulation is random If not set patches may still have individuals above carrying capacity after modifying the population Note that if do flush is not set but do fill is set patches are automatically regulated to be able to fill the empty undersaturated patches with any supernumerary individuals available in the population resize_keep patch matrix opt This array parameter 1D matrix specifies which patches must be kept when resizing a population Its length will set the number of patches in the popu lation after resizing The patches are numbered from 1 to patch_number and they are ordered as specified by the patch_capacity parameter The order of IDs specified here is kept patches may thus be reordered with this option as shown is the next example patch_capacity 115 10 5 10 1007 resize_
93. nk one of these two is replaced by the other usually following an alphabetical order As each parameter may appear only once in the init file each LCE must be given only one rank value Giving several values to a LCE will make it a sequential parameter The way to build the life cycle in the init file is to write the LCEs names given below followed by their rank number Here is an example see chapter 6 for more details breed i save_stats 2 save_files 3 disperse 4 selection 5 aging 6 This very simple life cycle starts with mating and breeding within the population that will generate a new offspring generation provided adults are present within patches The statistics are then recorded and the simulation data is saved at the 30 CHAPTER 4 LIFE CYCLE EVENTS 31 right generation Because the save_stats LCE is placed after breed the data on both the offspring and adult individuals can be recorded This wouldn t be the case if it was placed after aging where only the stats on the adults would be recorded for instance The disperse LCE then moves the offspring around according to the migration model chosen The offspring then experience a round of viability selection within their patches where their survival probability is determined by the phenotypic value of the viability trait they carry They are then moved to the adult age class previously emptied of its previous occupants from the previous generation by the aging LC
94. o run Nemo under Windows You may install CygWin from http www cygwin com as Nemo has been compiled using this environment this is the better option Or you can simply use the MS DOS terminal i e the command prompt The latest option is explained here as using CygWin is like using any nix environment see previous section So launch the command prompt cmd exe and cd to where you have installed Nemo Assuming you have the following files in your current working directory after downloading the right archive i e Nemo z y z Win binaries zip nemo2 x y exe cygwinit dll Nemo2 ini The cygwin1 dll file is required to run Nemo outside of CygWin and must be sitting in the same directory as the nemo executable To launch Nemo simply type the command CHAPTER 1 INTRODUCTION 10 gt nemo2 x y exe You should have the same output as previously under MacOS X Linux Note about CygWin when installing CygWin check that you also install the GSL library by checking the gs1 Runtime option under the Libs section of the installer 1 3 2 Batch mode Nemo accepts only one type of argument on the command line the name s of the init file s to run simulations from For instance if three init files are passed to Nemo the program will initiate three simulations from those files considering they don t incorporate any sequential parameters Sequential parameters are parameters with more than one argument value se
95. oci of the trait by the number of chromosomes or set by the parameter prefix _chromosome_num_locus below The random positions are saved in the log output file of the simulation and in the binary file as well CHAPTER 5 TRAITS 62 prefiz_recombination_rate decimal array opt This option lets one set the positions at equal distance between loci on a given chromosome A recombination rate of 0 01 corresponds to a map dis tance of 1 cM Therefore if smaller recombination rates are specified the map resolution will be reset accordingly The number of chromosomes is deduced from the number of elements of the array and the number of loci per chro mosome is either equal among chromosomes and set by dividing the number of loci of the trait by the number of chromosomes or set by the parameter prefiz_chromosome_num_locus below If a single value is given without using a matrix argument a single chromo some is constructed If a single value is given and that value is 0 5 the loci are considered as unlinked and recombination is handled independently of the genetic map Therefore if two traits have a recombination rate of 0 5 their loci will be considered as unlinked altogether This would however not happen if an array argument is passed e g with ntrl_recombination_rate 0 5 and delet_recombination_rate 0 5 in which case the loci of the traits will have same map positions although they are unlinked to the next loci pref
96. of the current simulation see replicates above each replicate will use a differ ent source file as a source population In the case this value is smaller than the current number of replicates the source population will be changed every replicates source_replicates replicates The source filename is built using the value of the source_pop parameter to which the replicate counter and the file extension are added Therefore the source_pop parameter string value must not include these character strings The replicate counter is built using the digit information given below by the source_replicate_digit parameter source replicate digit integer opt This parameter is needed build the replicate counter of the binary source file name when the parameter source_replicates is specified Its value must match the number of digits used in the replicate counter of the source file names For instance it is 3 if one of the source filenames ends with say 032 bin source_start_at_replicate integer opt The first replicate to load data from can be set using that parameter The rules described above to set the replicate number applies but start at the value set here rather than 1 Examples The first example shows how to load a population from the last generation saved in a single source file in preserve mode replicates 10 source_pop binarydir mysourcepop_001 bin source_preserve Here the same population from the file name
97. of physical maps The genetic map may be composed of more than one chromosome each with a different number of loci although not always see options below The recombination distances between loci can be specified explicitly or set randomly This way for instance neutral markers SNPs can be located more or less closely to loci under selection This is done thanks to a set of parameters that are common to the three traits and are described in this section 60 CHAPTER 5 TRAITS 61 The naming convention for the genetic map parameters is prefix_parameter_name where prefix stands for ntrl quanti dmi or delet The unit of the map is the centi Morgan cM by default but can be changed if needed with parameter prefix _genetic_map_resolution The map parameters are optional by default and unlinked maps for each traits will be built if no parameters are specified in input that is all loci are unlinked There are four types of maps fixed maps prefir_genetic_map which specify the exact map position of each locus on each chromosome random maps pre fix_random_genetic_map which randomly set map positions according to the map length of each chromosome fixed maps with equally spaced loci prefiz_recombination_rate which set locus positions according to specified recombination rates specific to each chromosome and trait and unlinked maps by default or if prefiz_recombination_rate 0 5 which correspond
98. of their fitness trait against a fitness function or aging will remove all adult individuals independently of their traits value to make room for the new generation The simulation components can also declare different output files and statistics The file extensions and stat outputs are indicated for each component For a discussion and a complete list of output statistics have a look at chapter 7 21 CHAPTER 3 SIMULATION COMPONENTS 22 3 1 Simulation name simulation files log stats NA replicates integer Number of replicates to perform per simulation generations integer Number of generations performed per replicate filename string This name will be used as the base filename of all output files of a simulation The output file extensions are added to this base filename by the different simulation components that write data to files If a file is written on a replicate periodic basis the replicate number will be added between the basename and the extension so that the same file is not overwritten periodically The same is true concerning generation periodic files see section 2 7 The base name may include the special expansion character used to build filenames when sequential parameters are present in the input parameter file See the discussion on sequential parameters in chapter 2 root_dir string opt The path specified by this parameter will be used as the root directory path for all ou
99. ompatible matings adlt fwoinf mean demic infection in extant demes for adult females adlt mwoinf mean infection frequency of adults males in the whole population wolb infvar inter demic variance in adult female infection wolb extrate proportion of demes having lost infection in adult females wolbachia _perpatch off pzfwoinf mean infection frequency of offspring females in patch i Table 7 8 Wolbachia stat options continued off pimwoinf mean infection frequency of offspring males in patch 2
100. ontrol for patch extinction The extinction rate is used as the probability of an event to occur for each patch be it total extinction or harvesting The rate harvesting size and harvesting proportion parameters can be set differently for each patch by using a matrix argument They will affect all age classes equally unless the harvesting size is drawn from a random distribution The sex of the individuals that are removed is set randomly extinction rate decimal matrix opt Probability per generation that a patch undergoes extinction or harvesting Defaults to 1 The default behavior if none other parameters are given is to completely empty the patch of all its individuals when an extinction event occurs extinction_size decimal matrix opt The number of individuals to be removed from a patch when the event occurs Alternatively the mean of the distribution of harvesting sizes see bellow extinction proportion decimal matrix opt The proportion of individuals to be removed from the patches in case of har vesting The size parameter has precedence over this one extinction threshold decimal opt CHAPTER 4 LIFE CYCLE EVENTS 44 The threshold is set as the minimum density of individuals relative to the patch carrying capacity that must be present in the patch to consider it as non extinct including all individuals in the patch offspring and adults If the patch density is below that threshold the patch is emptied ex
101. or one If only one initial trait value is specified per patch that same value will be used for all traits CHAPTER 4 LIFE CYCLE EVENTS 45 quanti_init_freq matrix opt Similarly the matrix must hold the patch specific allele frequencies row wise and locus specific frequencies column wise The frequency of the first allele only needs to be specified As said above the initialiser assumes there are only two alleles per locus see quanti trait parameters quanti_allele_model and quanti_allele value The same remarks hold concerning value recycling 4 9 2 Initialization of trait ntrl name ntrl init age flags unchanged files NA This LCE can be used to set initial allele frequencies in each patch differentially It assumes loci carry only two alleles ntrl_init_patch_freq matrix opt This is the same as for quanti_init_freq above although for the ntrl trait instead 4 9 3 Initialization of trait dmi name dmi_init age flags unchanged files NA This Life Cycle Event is used to set the frequencies of the mutant alleles at first generation It allows setting the frequencies in a patch wise manner The frequencies at first generations will match those specified here on average because they are used as probabilities to sample mutations within a deme In absence of an initializer all individuals are monomorphic for the wild type allele at all loci dmi init freq matrix A matrix with one row per patch and one column
102. ost decimal This is the probability that a dispersing offspring dies during dispersal The female and male costs are identical dispersal_cost_fem mal decimal opt CHAPTER 4 LIFE CYCLE EVENTS 39 These two parameters set the dispersal costs affecting male or female dispersers separately They will be overridden if the previous parameter is also present and they must be set together to set this LCE correctly dispersal fixed trait female male opt One of the sex dispersal gene can be turned off with this parameter The individuals of the selected sex will then migrate following the dispersal rate given below dispersal fixed rate decimal opt This is the dispersal rate of the non evolving sex 4 7 Selection name viability_selection integer age flags offspring required files NA derived components breed selection breed selection disperse Viability selection selectively removes individuals from a patch based on their sur vival probability given by their fitness trait Currently the fitness determining traits are delet deleterious mutations quant quantitative traits and dmi Dobzhansky Muller incompatibility loci although any other trait may be used as long as the trait s phenotype is compatible with the fitness models implemented Fitness can be either absolute i e directly set from the individual s phenotype or relative to the mean fitness value of the patch or of the whole population Fo
103. ot_dir test random_seed 988889 run_mode overwrite replicates 10 generations 1000 POPULATION patch_number 50 CHAPTER 6 EXAMPLES 81 patch_capacity 20 HH LIFE CYCLE breed_selection 1 save_stats 2 save_files 3 disperse_evoldisp 4 aging 5 store 6 extinction 7 breed and selection parameters selection_trait delet selection_model direct mating_system 3 monogamy mean_fecundity 15 high enough to resist inbreeding depression mating_proportion 0 8 20 of extra pair mating extinction parameter extinction_rate 0 05 disperse parameters dispersal_model 2 dispersal_propagule_prob 0 3 dispersal_rate 0 125 save_stats parameters stat off fstat off delet viability disp demography extrate stat_log_time 10 stat_dir stat store parameters store_dir binary store_generation 1000 store_noarchive NEUTRAL MARKERS ntrl_loci 20 ntrl_all 256 ntrl_mutation_rate 0 0001 ntrl_mutation_model 1 ouput ntrl_save_genotype CHAPTER 6 EXAMPLES 82 ntrl_output_dir ntrl ntrl_output_logtime 1000 GENETIC LOAD delet_loci 100 delet_init_freq O delet_mutation_rate 0 0001 delet_effects_distribution exponential delet_effects_mean 0 05 delet_dominance_mean 0 36 delet_fitness_model 1 ouput delet_save_genotype delet_genot_dir delet delet_genot_logtime 1000 DISPERSAL GENES disp_mutation_rate 0 001 disp_mutation_mean 0 2 dispersal_cost 0 2 This exampl
104. ow much characters must be read within the format string and added to the filename For instance for value no 4 of sequential parameter no 3 above the string Ba will be added to the filename string A last option is to replace the character string by a to replace the argument value by its position value 1 3 As here the third sequential parameter is supposed to have 6 argument values the stands for the integer values 1 to 6 and the width specifier is 1 no leading 0 Here is the full example filename a 4 1_b 3 2_ 2 AaAbAcBaBbBc 3 my_seq_param_1 1 10 1500 CHAPTER 2 THE INIT FILE 17 my_seq_param_2 0 001 0 01 0 1 my_seq_param_3 matrix no 1 matrix no 2 matrix no 6 These settings will give the following simulation filenames 54 total a0001_b001_Aa a0001_b001_Ab al500_b100_Bc The number of simulations initiated by sequential parameters is equal to the product of the number of arguments of each sequential parameter All the parameters value combinations are performed There is currently no way to restrict the number of combinations 2 5 External argument files It is sometimes convenient to write large matrices or large numbers of sequential parameter arguments in a separate text file and only specify the path to such file s in the init file This is done by providing the path to the file with the filename syntax where filename is a character string that contains
105. p stat option is given ERROR the string disp is not a valid stat option ERROR could not run the sim 7 1 Stat Output Files The save_stats LCE declares two output files the txt and _bygen txt files The first filetype contains the stat records of each recorded generation set with the stat_log time parameter for each replicate By default the first and last generations 83 CHAPTER 7 OUTPUT STATISTICS 84 are automatically recorded This file may be huge depending on the number of stats you are monitoring It adds two columns the replicate and the generation columns containing the replicate number and the generation number respectively The _bygen txt file only contains the generation column as each line contains the stats averages taken over all replicates One extra stat is added alive repl it counts the number of extant replicates at each generation The replicate stats are dumped to the txt file at the end of each replicate whereas the stat average values are saved to the _bygen txt file at the end of a simulation 7 2 Stat Options The following tables present the different summary statistics of the simulation com ponents that can be monitored during a simulation run Output names beginning with off are computed on the offspring age class while those starting with ad1t are computed on the adults When a stat is described as being the mean of a particular value this stat is the a
106. per locus specifying the initial allele frequency at each locus in each patch Both the number of rows and the number of columns can be smaller than the actual number of patches and loci respectively If so the pattern present in the matrix will be repeated over all patches loci Examples with 6 demes and 8 loci CHAPTER 4 LIFE CYCLE EVENTS 46 to set all loci in all demes to allele i dmi_init_freq 1 to set the allele frequency to 0 25 in every second deme dmi_init_freq 0 25 0 to set loci 1 2 5 6 to allele 1 in demes 1 3 5 and to allele O at the other loci in the other demes dmi_inti_freq 1 1 0 0 0 0 1 1 same as above but with explicit repetition of the pattern of frequencies over loci dmi_inti_freq 1 1 0 0 1 1 0 0 0 0 1 1 0 0 1 1 dmi_init_patch matrix opt This optional parameter allows to restrict the settings given above to a speci fied set of demes This is usefull to set allele frequencies is some demes only Will have an effect on gene dynamics under stepping stone lattice dispersal only Example with 6 demes and 8 loci to set one patch with all loci to allele 1 dmi_init_freq 1 dmi_init_patch 6 this is patch no 6 to set three first patches to allele 1 at all loci dmi_init_freq 1 dmi_init_patch 1 2 3 etc Note that this would be equivalent to setting the frequencies in each deme explicitly This option is a shortc
107. portion of within demes selfed matings meanviab off viab see above adlt viab same for adults Table 7 4 Deleterious mutations continued on next page CHAPTER 7 OUTPUT STATISTICS 93 Stat option Output name Description survival __ now part of the selection LOL stats Table 7 4 Deleterious mutations stat options continued 7 7 Dobzhansky Muller Incompatibilities DMI Table 7 5 DMI stat options Stat option Output name Description adlt off dmi age dmi freq overall average frequency of mutant alleles across loci age dmi pi patch specific frequency of mutant alleles across loci age dmi icmp overall average frequency of the incompatible genotype s AaBb for diploids Ab and aB for haploids age dmi icmp pi patch specific frequency of the incompatible genotype s across loci Table 7 5 DMI stat options continued 7 8 Selection Table 7 6 Selection stat options Stat option Output name Description fitness age fitness mean mean of the within patch offspring fitness before viability selection i e including all offspring fitness outb fitness of b n demes outbred offspring fitness outw fitness of w n demes outbred offspring fitness hsib fitness of half sib crosses fitness fsib fitness of full sib crosses fitness self fitness of selfed crosses fitness prop prop outb proportion of b n demes outbred offspring
108. r opt Changes the patch carrying capacity for the males only similar to pop_nbmal resize_age_class offspring adults all opt Sets the age class of the individuals to use when filling up new or empty patches If no individuals of the required age class are present in the popula tion the LCE does not modify the population It defaults to all resize_do_flush bool opt This parameter tells what to do with supernumerary individuals that are pro duced when patches are removed from the population It also conditions the way patches are filled CHAPTER 4 LIFE CYCLE EVENTS 48 When set present any supernumerary individuals will be flushed removed and patches may subsequently be filled using individuals created de novo i e they are similar to first generation individuals and have no parents When not set absent supernumerary individuals are backed up and may then be used to fill the remaining patches This option is necessary when simulating patch fusion e g bring the individuals from two patches into one or fission e g create two patches from one resize_do_fill bool opt If set the patches will be filled after the patch number and or the patch carrying capacities have been modified The individuals used to fill the patches are either backed up individuals i e do flush is not set or first generation individuals i e do flush is set see comment above Patches will be filled sequentially starti
109. r now it only acts on the offspring age class but can be placed anywhere in the life cycle Future releases will extend this behaviour to selection on other age classes This LCE also declares a set of fitness statistics that can be recorded during the simulation see section 7 2 The parameters described here are the same as those used with the breed_selection and breed selection disperse composite LCEs which inherit those parameters New in 2 2 selection can now act on multiple traits simultaneously That is the fitness of an individual is given by the multiplication of the fitness values provided by each trait under selection See section 4 7 1 below selection_trait string The argument to this parameter must be the name of the trait under selec tion Only one trait can be specified would become a sequential parameter otherwise The traits name are found in the next section Currently the delet quant and dmi traits are the only traits under viability selection i e their trait value is used to set the individual fitness CHAPTER 4 LIFE CYCLE EVENTS 40 selection_model fix direct gaussian opt The selection models are fix The fitness of the individual is set according to its pedigree and the number of lethal equivalents The model used here is the following Wp Woxe7 where WF is the fitness of an individual with pedigree inbreeding coefficient F Wo is the base fitness of the population set below and A is
110. r specifies a specific path used to save the genotype output files Should not end with a slash delet_genot_logtime integer opt This is the generation periodicity of the genotype files or the generations at which the files should be saved if provided as multiple values in an array If CHAPTER 5 TRAITS 74 the number is greater than the total number of generations no data will be saved 5 5 Dobzhansky Muller Incompatibility loci name dmi files dmi phenotype a real value in 0 1 interpreted as the fitness value of the individual The DMI trait codes for so called Bateson Dobzhansky Muller Incompatibilities that occur between pairs of loci when both loci are heterozygotes for diploids or for heterozygous pairs for haploids In the latter case loci in repulsion usually decrease fitness i e aB or Ab have lower fitness than AB or ab The trait is bi allelic with allele 0 representing the wild type A B C and 1 the mutant a B C The fitness effects of each incompatible pair must be set using a matrix argument see dmi_genot table below The fitness values of all possible genotypes must be specified The fitness model used is multiplicative W a w pair where w pair is the fitness value of the locus pair i A specific initializer has also been added to set patch specific initial frequencies see dmi_init dmi loci integer Number of incompatible locus The trait is
111. s and dams are randomly selected with or without replacement within each patch depending on the value of the cross_with_replacement parameter cross_num sire integer Number of sampled males per patch Each male will be mated with num_dam females as many times as num_offspring cross_num_dam integer Number of sampled females per sire Each female produces num_offspring with one given male cross_num_offspring integer Number of offspring produced per dam cross_at_generation integer Generation at which crossing is performed cross_do_within_pop bool opt If set the default dams and sires will be sampled within populations cross_do_among_pop bool opt If set the crossings will be performed by sampling a sire and a dam from two different populations Sampling proceeds by first randomly selecting num_sire males within each patch and randomly assigning num_dam females to each sire taken from patches different from the sire s one This insures that the sire and the dam of each cross are from a different patches Both within and among patch crosses can be performed if both options are set CHAPTER 4 LIFE CYCLE EVENTS 51 cross_with_replacement bool opt If set to 1 true this option allows to sample individuals with replacement that is to sample several times the same individual when selecting dams or sires for the crossings If not present or set to 0 the sampling is done without replacement which
112. sizes matrix e explicit pairwise dispersal rates also sex specific dispersal matrix e demographic stochasticity built in e extrinsic extinction rate or patch size variation harvesting e temporal change of the population parameters and or dispersal rates e pure Wright Fisher population model with constant population sizes with the breed disperse LCE 1 2 2 The Individual An individual in Nemo is basically defined as a trait container That means that the phenotypes of the individuals depend on which traits are modeled based on the parameters in the input file By default individuals don t carry any genetic informa tion in absence of traits The only pre defined phenotypes are the individual s age and sex Individuals also store information about their ancestry and demographics and have a unique ID and a pedigree class informs if the two parents were a single in dividual full sib half sib or unrelated individuals Each individual also stores the CHAPTER 1 INTRODUCTION 6 number of babies it had and the ID number of its mum dad and natal patch These information tags are used to compute pedigree based or age sex specific statistics and are sometimes saved to file by the different simulation components 1 23 Genetics The genetic models implemented depend on the type of traits but all types of loci can be placed on the same genetic map That map is a recombination map and not a physical map in that it specifies the
113. smission is vertical through females only and is not perfect the zygote may loose its parasite mutation process rep resented by the transmission rate presented below Zygotes issued from the mating between an infected male and an uninfected female must pay the cost of incompati bility that decreased their chance of survival at birth by a given amount parameter incompatibility cost of the breed wolbachia LCE Being infected by Wolbachia also induces a cost that translates into a reduced fecundity of the infected females parameter fecundity cost of the breed_wolbachia LCE See the breed_wolbachia LCE for details on the breeding and infection parameters wolbachia_transmission_rate decimal This is the rate of transmission of Wolbachia from a mother to its offspring If different from one the parasite may be lost during gamete formation Chapter 6 Examples 6 1 Life cycles 6 1 1 A basic life cycle To start with lets exemplify what a basic life cycle looks like breed 1 disperse 2 aging 3 It starts with the reproduction of the population breed thus adding offspring individuals to it Then the offspring migrate within the population disperse before getting older and replacing the previous adult generation that will die because of aging non overlapping generations The new adult generation is also regulated to not exceed the patches carrying capacities Writing this life cycle in a different order wou
114. t all a number will be added to the simulation filename to prevent overwriting the same file s several times does not apply to other string arguments The simulation base filename will get an extra extension of the form at its very end where stands for the number of the simulation in the sequence ex if we have these two sequential parameters patch_number 10 50 patch_capacity 5 10 Setting the base filename this way filename 2pop_ 1ind source_pop 2pop mysource_ 1ind will give the following basenames one for each simulation 10pop_5ind 50pop_5ind 10pop_10ind 50pop_10ind Here 2 refers to patch_number and 1 refers to patch_capacity in alphabetical order If the filename parameter is specified without expansion character CHAPTER 2 THE INIT FILE 16 filename mysim the simulation basenames will be mysim 1 mysim 2 mysim 3 and mysim 4 Advanced filename expansion The system presented above works fine when the sequential arguments are numbers even floating point numbers that can easily fit into a filename string However when for instance the sequential argument is a matrix or is too long to fit in we also want to have a way to get a specific filename that we can refer to more explicitly than by a number This is done by adding a format string within the expansion string That string helps setting the format of the argument value number of digits to use or provides an a
115. tch_number x patch_number in dimensions Each d element of this matrix is the dispersal probability from patch 2 to patch J This parameter has precedence over the dispersal rate and model parameters If too big and especially when containing a large number of zeros can be replaced by the dispersal_reduced_matrix and dispersal_connectivity_matrix below dispersal matrix fem mal matrix opt The dispersal matrices are in fact sex specific and this parameter can thus be used to specify sex specific dispersal patterns Same comment about the precedence as above dispersal_rate decimal opt This parameter sets both the male and female dispersal rates identical value for both Nemo will build the dispersal matrices according to the dispersal model chosen dispersal rate fem mal decimal opt Replaces the previous parameter for the case of different males and females dispersal capabilities dispersal reduced matrix matrix opt This matrix holds the non zero dispersal rates from patch i row wise to patch j column wise where the identity of the connected patch j is provided by the dispersal_connectivity_matrix parameter see below Because not all patches may be similarly connected to other patches the number of elements per row may vary For each row focal patch the number of elements must exactly be the same as in the dispersal connectivity_matrix The sum of each row must be one CHAPTER 4 LIFE CYCLE EVENTS 3
116. ted fst within age f st1 deme specific Far within deme 2 for all demes lad1t off fstwC age fis WC the Weir amp Cockerham 1984 Fis estimate f age fst WC the Weir amp Cockerham 1984 Fr estimate 0 age fit WC the Weir amp Cockerham 1984 Fir estimate F lad1t off age D mean b n demes Nei s genetic distance mean NeiDistance D lad1t off age Di j pairwise Nei s genetic distance b n NeiDistance demes and J for all pairs adlt off Dxy age Dxy average pairwise sequence divergence between all pairs of patches adlt off Dxy patch age Dxy pip7 average pairwise sequence divergence between patch and patch j Table 7 2 Neutral markers stat options continued 7 5 Quantitative traits Table 7 3 Quantitative traits stat options Stat option Output name Description lad1t off age qi quanti mean phenotypic value of the trait in the whole population equal to the average breeding value in case no environmental variance is set age qi Va average of the within patch additive genetic variance Va of the trait age qi Vb among patch genetic variance Vb of the trait variance of the patch means Table 7 3 Quantitative traits continued on next page CHAPTER 7 OUTPUT STATISTICS 90 Stat option Output name Description age qi Vp average of the within patch phenotypic variance Vp present only
117. termine the content of output files especially the stats output files the ranks of these LCEs are thus important in that perspective A third output LCE could have been added here it is the store LCE Its rank in the life cycle will also determine the age class content of the binary files 6 2 Traits To add a trait to a simulation it is sufficient to add the mandatory parameters of that trait to the init file Here is an example with three of the traits currently implemented in Nemo CHAPTER 6 EXAMPLES 80 HH NEUTRAL MARKERS ntrl_loci 20 ntrl_all 20 ntrl_mutation_rate 0 0001 ntrl_mutation_model 1 SSM model GENETIC LOAD delet_loci 1000 delet_mutation_rate 0 0001 delet_effects_mean 0 05 delet_dominance_mean 0 36 delet_fitness_model 1 multiplicative model DISPERSAL GENES disp_mutation_rate 0 001 disp_mutation_mean 0 2 Each individuals in the simulation will thus carry four sets of genes One coding for neutral markers with 20 loci one with 1000 loci carrying deleterious mutations and two coding for female and male dispersal The genotypes can be saved in binary files using the store LCE or by adding the trait specific output parameters and the save files LCE somewhere in the life cycle 6 3 A complete example The next example shows a complete init files with all the mandatory parameters and all the trait output parameters SIMULATION filename example logfile logfile log ro
118. the path to the external file relative to the directory from which Nemo is run More than one external file can be provided in argument to a parameter in which case the parameter becomes a sequential parameter The expansion character can also be used in the filename character string NOTE the external file must be terminated by an empty line Otherwise it just needs to hold the argument s of a given parameter in exactly the same way as it would be written in the init file i e without new lines between multiple arguments Example param0 1 2 3 parami Ofilenamel txt filename2 txt Ofilename3 txt param2 path 1 to filename 1 abc 2 txt Here paraml and param2 have argument values stored in external files The filename and the directory path to param2 depend on the argument value of param0 and paraml i e path 1 to filename a txt path 1 to filename b txt etc CHAPTER 2 THE INIT FILE 18 2 6 Temporal arguments Nemo offers the possibility to change the value of a parameter during the course of a simulation and thus to modify the state of the population or of any particular component during a simulation Temporal arguments are limited to the non trait components for now They are specified in the init file by using the temporal argument specifier g within the argument string where the stands for the generation at which the argument value has to be used The state of the compo nents that have temporal
119. tinction size distribution uniform poisson normal exponential log normal opt The distribution used to randomly draw the harvesting size of a patch The mean of the distribution is taken from the extinction_size parameter In case of the normal and lognormal distributions the standard deviation of the dis tribution must be specified with the parameter below The harvesting size is drawn from the distribution for each age class separately i e offspring and adults extinction_size_dist_stdev decimal opt The standard deviation of the normal and lognormal random distributions for harvesting sizes 4 9 Trait initialization Patch specific trait or allelic values cannot be specified with the trait parameters Instead we need to use an LCE to perform this task Such LCEs are implemented for the quant dmi and ntrl traits 4 9 1 Initialization of trait quant name quanti_init age flags unchanged files NA There are two possibilities to initiate the quantitative trait one by specifying the mean trait value in each patch and the other by specifying the mean allele fre quencies per locus The allele frequency initialisation is performed for bi allelic loci only quanti init trait values matrix opt The matrix must hold patch specific trait values in each row If the number of rows is lower than the number of patches values will be recycled The number of values per row must either the same number of traits modelled
120. tion rate The SSM model 1 changes the existing allele number k to the k 1 or k 1 value randomly The boundaries are reflexives the allelic value can not exceed the ntrl_all value or be less than 1 The KAM model 42 modifies the existing allele by assigning it a new random value within the LO ntrl_all range CHAPTER 5 TRAITS 64 ntrl_init_model 0 1 opt This option sets the way marker genes are initialised The mode 0 means no variance all alleles have same value i e 0 at the start of a replicate Mode 1 means maximum variance the allele values are set randomly within the range 1 ntrl_all Mode 1 is the default mode See section 4 9 2 for a different way of initialising allele frequencies within patches ntrl_recombination_rate ntrl_genetic_map ntrl_random_genetic_map opt Recombination is handled by the genetic map All genetic map parameters apply See section 5 1 ntrl_save genotype string opt If this parameter is present the population genotypes will be saved in a text file with the dat extension Three file formats are proposed depending on the argument passed to this parameter capital or non capital letters are accepted e TAB tab The allelic values are saved on one line per individual and two columns per locus This format is ideal for the R software and analysis with the HIERFSTAT R package by J Goudet e FSTAT fstat The file format is almost the same as that used
121. tivariate Normal distribution can be correlated In this way the evolution of correlated traits and genetic constraints on adaptation can be modeled Environmental variance can also be modeled as well as spatially and temporally varying selection pressures see the selection LCE The statistics implemented return the additive genetic variation within populations Va the among populations genetic variance Vp the Qgr index of trait differenti ation among populations Qsr r and the traits genetic correlation along with the eigenvalues and eigenvectors of the G matrix within demes or the D matrix among demes when two or more traits are modelled quanti_traits integer The number of traits to model The number of traits is not limited If two or more traits are modelled the mutational covariance can be set and the statistics returned include the genetic correlation of the traits and the eigen decomposition of the genetic covariance matrix both within G matrix and among D matrix demes quanti_loci integer Number of additive loci that determine the trait s Loci are diploid The trait value is set by summing the allelic values at all loci When two or more traits are modelled they share the same loci and each locus has an effect on each trait 1 e fully pleiotropic loci The mutation effects on the traits can be more of less correlated depending on the mutational covariance see below quanti_mutation_rate double
122. to completely unlinked loci The map resolution that is the minimum distance at which crossing over will be placed depends on the mini mum resolution specified by the map parameters of the different traits and can be explicitly set by prefiz_genetic_map_resolution Limitations are that the number of chromosomes can not differ among traits i e chromosomes without loci are not accepted and the number of loci per chromosome on fixed map must be constant see below prefiz_genetic_map matrix opt This corresponds to a fixed map and is used to specify the map position of each locus of a trait The matrix argument provides the locus positions using one line per chromosome in cM by default The number of chromosomes is then deduced from the number of lines Note because matrices in input must carry the same number of elements per line this parameter does not allow for different number of loci per chromosome This is not true for the other types of map prefiz_random_genetic_map array opt Loci position can be set randomly on the map Here the array holds the map size of each chromosome in cM by default The number of chromosomes is deduced from the length of the array and the loci positions are drawn ran domly from a uniform distribution on the range 0 map size By chance two loci may land on the same map position The number of loci per chro mosome is either equal among chromosomes and set by dividing the number of l
123. tput files and directories declared by the simulation components This path will thus be added in the front of any other paths defined subsequently e g by param stat_dir run mode string opt This sets the simulation behavior with the following options overwrite previously saved files with the same base filename as the current one are overwritten A warning is issued on the standard output i e terminal window dryrun does not run the simulation but just sets the parameters and checks for the files and statistics The output paths and log files are created create_init similar to dryrun but writes the parameters of each possible simulation in a separate init file in the working directory This is handy when wishing to create many init files from a single one containing many sequential parameters CHAPTER 3 SIMULATION COMPONENTS 23 skip automatically skips simulations whose base filename already exists on disk run default the default running mode silent_run turns off all regular and warning messages only the error mes sages are issued logfile string opt This is the file in which the simulation logs are recorded The simulation basename and each directory paths are recorded as well as the mean elapsed times for the simulation and the replicates and the dates of beginning and end of a simulation By default Nemo will save all this information in a file named nemo log in its working dir
124. ues with as many columns as the number of trait pairs or just one value if the correlation is meant to be identical for all trait pairs selection trait dimension integer opt Sets how many dimensions or quantitative traits are modeled selection _local_optima matrix opt A single array of local phenotypic optima for each quantitative trait or a matrix with at most as many rows as the number of patches to set the patch specific optimum values for each trait The spatially explicit matrix is dealt with in the same way as for the selection variances and correlations selection rate environmental change decimal array opt A single decimal number interpreted as the rate of change of the optimum phenotypic values in all patches and for all traits or an array of trait specific rates of change of the phenotypic optima in all patches The array may contain less values than the number of traits in which cases the values are recycled among traits The rates are here absolute rates For instance a rate of 0 1 will change the local phenotypic optima by 0 1 units per generation e g 3 gt 3 1 gt 3 2 gt 3 3 gt 3 4 etc This rate is thus independent of the amount of genetic variation in a population This can be changed by using the set of parameters below selection_std_rate_environmental_change decimal array opt Same as above to the difference that the rates are interpreted as unit of phe notypic standard deviation The
125. ut when the number of demes is large e g this would be equivalent to the two examples above dmi_init_freq 0 0 0 0 0O 1 set in patch 6 only dmi_init_freq 1 1 1 O O 0 set in patch 1 2 and 3 CHAPTER 4 LIFE CYCLE EVENTS 47 4 10 Resize Population name resize integer age flags unchanged files NA The resize LCE modifies the state of the meta population during a simulation but with more control than by using temporal arguments within the population param eters In particular it allows the user to merge or split existing patches without losing individuals or adding empty patches which is what would happen when us ing temporal parameters resize_at_generation integer matrix This is the generation at which the population will be modified Mandatory This parameter also accepts a matrix argument with all the generation num bers specified on a row Temporal arguments at the other resize parameters then allows the modification of the population state at different points during a simulation see examples below resize_patch number integer opt Specifies the new number of patches in the population resize patch capacity integer opt Specifies the new patch carrying capacity also accepts a matrix argument as the population parameter see 3 2 resize female capacity integer opt Changes the patch carrying capacity for the females only similar to pop_nbfem resize_male_capacity intege
126. vailable in Nemo replicates 1 generations 1 patch_number 5 patch_capacity 50 source_pop source path srce fstat file dat source_preserve source_file_type dat source_fill_age_class adults LIFE CYCLE save_stats 1 CHAPTER 3 SIMULATION COMPONENTS save_files 2 stat adlt fstat adlt fstWC adlt weighted fst stat_log_time 1 stat_dir stat NEUTRAL MARKERS ntrl_loci 20 must match the number of loci in the file ntrl_all 10 same for the number of alleles ntrl_mutation_rate 0 useless here but mandatory parameter 29 Chapter 4 Life Cycle Events The life cycle events hereafter LCE are operators used to modify the state of the population and interact with the different components of a simulation Each LCE is executed only once during the course of a generation at the rank it has been assigned in the stack of LCEs that constitutes the life cycle This rank is given by the user in the init file The life cycle is thus an ordered list of LCEs selected by the user Most LCEs act on a per generation basis Some may however have a different periodicity set by the parameters they declare The ranks should start with value one for the first LCE and be incremented for each successive LCE As the LCEs are placed in ascending order in the life cycle their exact rank value does not matter so much as long as the order is conserved i e the rank increment may be different from one If two LCEs have same ra
127. verage of the patch means of the value Some stat options may take a prefix tag specifying on which age class they are computed The naming convention is as follows A stat argument specified as lad1t off name has three possible forms adlt name off name or name mean ing the statistics can be restricted to one of the two age classes or computed for both Alternatively a stat option described as ad1t off name has only two forms adlt name or off name Likewise a stat option without any age class prefix does not accept any such option and likely apply to all age classes unless specified oth erwise Table comment Stat option the argument of the stat parameter in the input file Output name the name of the stats as written in the output files CHAPTER 7 OUTPUT STATISTICS 85 7 3 Population Table 7 1 Population stat options Stat option Output name Description off demography off nbr total number of offspring in the metapopulation off nbfem mean number of female offspring per extant patch off nbmal mean number of male offspring per extant patch off density average offspring density off dvar variance of the offspring density of extant patches adlt demography adlt nbr total number of adults in the metapopulation adlt nbfem mean number of females per extant patch adlt nbmal mean number of males per extant patch adlt density average adult density adlt dvar

User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents