Home

View

1. Pigblup Clinic August 1993 29 The environmental trends depict the change of conditions under which animals have had to perform over time The genetic and environmental trends together give an indication of the average production level at various times A decrease in the environmental trend could be caused by the animals having endured a deleterious environmental change eg a severe heat spell or a disease outbreak or could be due to there having been too few animals in the group for reliable estimation A corollary is that an increase in the trend could also arise from poor estimation This highlights the need for the breeder to take account of these factors when interpreting his output Possible Future Short Term Enhancements It is hoped that interaction with participants at this clinic will give the PIGBLUP team other suggestions for possible future enhancements and upgrades The following are some potential short term enhancements suggested by the developers and the PIGBLUP Consultative Group Introduced Animal groups address the problem which arises when animals from outside the herd are introduced for breeding at various times throughout the breeding program Current versions treat these animals as though they come from the same genetic base when in all likelihood they would be more superior to the base animals the more recently they were introduced The effect of this limitation is that those animals and their progeny are penalised i
2. animals o Selection differentials for traits enable the breeder to assess the relative selection emphasis and intensity for analysed traits The average generation interval for boars and sows provides an indication of the rate at which animals are being turned over Average annual inbreeding provides a handle on this aspect of his breeding program 3 7 Mate Selection Since BLUP uses information from all known relatives there 1s an increased risk of inbreeding in the herd especially when a trait of low heritability is being selected Inbreeding can be ameliorated by increasing the number of boars used and including production traits in addition to reproduction traits in the breeding objective PIGBLUP has a Mate Selection module which can be used to suggest matings in a particular week This module uses linear programming to maximise the average EBV s of progeny while minimising their average inbreeding coefficients The breeder can set the relative emphasis given to these two effects according to his her concern 3 8 Output The stages described above generate a large volume of useful information for the breeder Graphs of genetic and environmental trends allow the breeder to gain a perspective on how well his breeding program is progressing The genetic trends indicate how each EBV has changed relative to that of the base animals In addition the relative genetic mean for the herd at various times can be ascertained AGBU
3. on line printer At present you must wait for the printer to finish printing before PIGBLUP will allow you to do other things Users can now define the maximum size of a user defined management group Number Born Alive can now be selected or de selected for analysis Previously it was always selected 3 4 Running PIGBLUP PIGBLUP works by having the Menu program PBLUP cascade through each of the analytical modules as it progresses The menu program then waits for the launched module to complete check whether conditions are satisfied for moving to the next stage of the analysis and take the appropriate steps AGBU Pigblup Clinic August 1993 25 PPREP l PBMM2 LICOEFF3 SOLV INDEX2 AU4 MAT3 Figure 2 PIGBLUP Analytic Modules At present when you chose the analysis date prior to running a PIGBLUP analysis PIGBLUP resets the system date to that value This situation will be remedied in the next release of PIGBLUP When you choose the analysis date it will be written to a file which is passed with other information to the analytic modules This change will have the added benefit of the date being retained between sessions As most users use today s date when they run PIGBLUP today will be a valid date whenever you run PIGBLUP the current analysis date will be automatically used for the analysis Fundamental to use of BLUP is the requirement that we group animals which have performed under the same or similar c
4. What s New in PIGBLUP Tony Henzell 1 Introduction PIGBLUP is a microcomputer based system using the BLUP methodology to assess the genetic merit and extent of environmental influences on the performance of individual pigs The Estimated Breeding Values EBV s generated by BLUP for the traits chosen by the breeder for analysis are then combined by PIGBLUP to form a single EBV for each animal This EBV can be used as an overall measure of the animal s worth to the breeder for the market he is targeting With Version 2 1 of PLGBLUP a Genetic Audit module has been added to provide the breeder with breeding program diagnostics This module assesses how effective the breeder s selection practices have been in maximising genetic gain towards the defined breeding objective In addition to the breeder being able to monitor the genetic and environmental trends of the breeding program a Mate Selection module is also available now to calculate suggested matings which balance maximising genetic gain with minimising the effects of inbreeding on the herd To assist the user in understanding PIGBLUP s data requirements determine inputs drive the system and make better use of the output this paper gives an overview of how PIGBLUP implements BLUP technology the PIGBLUP menu system and the analytic modules it supports Enhancements and upgrades to various components of the system will then be more understandable and users will be able to get maximu
5. e for many of the sows Extra conditions to address this shortcoming should be investigated 21 day litter weight will be added as a selectable trait for analysis The restriction to a maximum of 520 weeks of data should be relaxed or if possible removed Also it should be possible to choose to analyse only the last n years of data in a PIGBLUP run Depending on the demand from breeders a Windows compatible version of PIGBLUP could be developed AGBU Pigblup Clinic August 1993 31 32 NOTES AGBU Pigblup Clinic August 1993
6. es responsibility from the user for management of the files for a number of herds breeds This avoids possible confusion as to which files pertain to which herd breed and to protect against accidental deletion of files It provides a degree of protection against input of inappropriate values and settings which might otherwise result in irrelevant analysis and wasted time It monitors each stage of the analysis with a view to determining whether conditions are satisfied for initiation of the next stage of analysis It provides an environment in which to view graph and or print results from the analyses Users of versions prior to Version 2 1 will find in the new version e The Menu system is more robust Better error checking and trapping now catches most errors and allows their cause to be identified more reliably than previously e Confirmation is now sought when file or directory deletion is requested and when the user tries to ESCape from the system Emphasis has been given to making operation consistent over all menus so that it is easier to leam and is more predictable in its effects Some functions have been stream lined to make editing or viewing easier this is most apparent in Mate Selection Scrolling both forwards and backwards is now possible in most menus in which files or long lists are examined Printing is now possible from within PIGBLUP to either a disk file for later printing or to an
7. exceeded PIGBLUP uses data on performance management groups and relatives together with information on the correlations between various traits to calculate EBV s PIGBLUP does this in two separate runs one for production and one for reproduction This is done for computing ease and assumes that production and reproductive traits are not correlated In other words knowing how many pigs litter a sow has gives little or no information on her genetic potential for growth or carcase characteristics A major advantage of BLUP 1s that it permits the EBV s for all selected traits to be computed simultaneously As a result all genetic and environmental effects are estimated more accurately The EBV s calculated in this way are the most accurate currently available As the EBV s of sire and dam may be combined to estimate the genetic merit of their progeny there is an immediate benefit from this increased accuracy BLUP solves the mixed model equations by making an initial guess at the solution and adjusting the solution at each iteration so that it moves towards the correct answer The larger the set of equations to be solved the slower this process 1s If the solution has not converged within the allocated number of iterations default 600 you have a choice of either increasing the number of iterations allowed or reducing the correction factors that should be used At each iteration an adjustment 1s AGBU Pigblup Clinic August 1993 27 c
8. lso been introduced to assess the breeding program the Genetic Audit and to recommend matings to the breeder the Mate Selection module PIGBLUP Version 2 1 3 1 Data Requirements The data submitted to PIGBLUP by the breeder s herd recording system needs to be formatted and sorted properly and all possible checks made so that the data is valid Users are reminded see the User Manual that PIGBLUP requires that animals are uniquely identified are sorted by date of birth and sorted by parity within animal As breeders use a number of different information management systems e g Herman MIPS Pigmania Pigchamp AIMS Pigman TAPS amp Fannplan a good deal of effort has gone into working with these software groups in developing their AGBU Pigblup Clinic August 1993 23 interfaces to the PIGBLUP system Although the final goal is the same for each of these programmers namely the PIGBLUP format they each have had to do things somewhat differently to get there as file structures will invariably be different for the different recording systems A standalone data checking program PIGCHECK which carries out a thorough check of the submitted data is being developed Suspect values are flagged in addition to outright errors and are written to an error log The user will be able to check data files using this program before trying to analyse data using PIGBLUP Potentially this will save user time and the need to contact ABRI f
9. m benefit from their system Finally some proposed future additions to the system will be suggested for consideration 2 Previous PIGBLUP Versions 2 1 Version 1 0 Version 1 0 of PLGBLUP was released in 1989 and calculated EBV s on just three traits ADG Lifetime Average Daily Gain BF Back Fat NBA Number Born Alive Calculated using a single trait repeatability model 22 AGBU Pigblup Clinic August 1993 2 2 Version 1 1 Version 1 1 was introduced in 1990 with two more traits available for analysis FCR Feed Conversion Ratio and TADG Average Daily Gain over a test period 2 3 Version 2 1 The most recent release Sept 1992 is Version 2 1 which expanded the set of available traits with o CFAT Carcase Fat CMUS Carcase Muscle and o LMY Lean Meat The breeder is restricted to choosing at most two of these three traits in an analysis System resources needed for computing increase with the square of the number of traits being analysed In order to analyse an increased number of traits and animals within the PC environment an improvement in the method of setting up of the mixed model equations in the computer was necessary AGBU developed the implicit representation of the mixed model equations to achieve this goal Benchmarking runs of this algorithm show that the new representation uses only 14 of the memory previously required at the cost of a 5 increase in run time Tools have a
10. n the estimation of their EBV s An option is being developed to enable these animals to be grouped according to period of introduction Improved EBV estimates should result The present INDEX is non linear and based on profit equations It appears as a black box This makes extrapolation of trends by the breeder more difficult A desired gains approach would give breeders more flexibility as to the amount of selection emphasis put on each trait Actual weightings used in the calculation of the index should be able to be reviewed by the breeder It is desirable that a breeder be able to restrict change to a genetic trait should he desire Introduce a pricing method for lean meat yield so carcase EBV s can be incorporated in the INDEX Presently a maximum of 900 EBV s are able to be sorted This could be increased to about 4000 and in addition allow any trait to be selected as the one on which to sort output 30 AGBU Pigblup Clinic August 1993 Currently the vertical scale on trend graphs 1s calculated automatically The ability to fix scale limits may be desirable in order to draw attention to the amount of gain achieved rather than just its trend Also we could provide greater flexibility in the defining of years to be considered in the Genetic Audit The Mate Selection module currently suggests alternative sires for mating should the primary sire be rejected for some reason Frequently the suggested alternative sires are the sam
11. occur within the context of a target market PIGBLUP uses two sub objectives a growing finishing sub objective and a sow sub objective when calculating marginal values for each trait The growing finishing sub objective takes account of economic and production information to calculate pertinent costs and returns for this aspect of the enterprise Marginal values for each production trait may then be calculated Similarly the sow sub objective uses the supplied costs to derive the marginal value of reproductive traits from producing an extra pig INDEX finally combines these sub objectives to calculate the EBV This index can be used by the breeder as an objective measure on which to base selection and culling decisions PIGBLUP does not remove the decision making from the breeder it is a tool for use by the breeder in decision making Note that at present carcase traits are not used in the EBV calculation 28 AGBU Pigblup Clinic August 1993 3 6 Genetic Audit Having calculated the trait EBV s and the EBV PIGBLUP performs a genetic audit to aid the breeder in determining the effectiveness of his breeding program Various measures are calculated The percent of animals selected versus those tested for both the current year and the 5 previous years provides a measure of the selection intensity of his breeding program No account is taken of pre selection occurring through the breeder choosing to omit measurements on some
12. omputed for each term in the solution If the correction factor is one the whole of the adjustment is added to the term in the previous solution A correction factor greater than zero but less than one 0 9 say says that only that proportion of the computed adjustment is to be added 0 9 of the adjustment say The effect of reducing the correction factor is to take smaller but surer steps at the penalty of needing more iterations to reach convergence Presently you can use an editor to alter the correction factors held in the file SOL V PAR but you are unable to alter the maximum number of iterations allowed The next version will make both factors editable from the Run PIGBLUP menu After each module has completed version 2 1 checks for conditions that would invalidate further processing before commencing the next stage in the analysis This helps to avoid wasting time 3 5 INDEX Once the EBV s for all selected traits have been computed for each animal the breeder is faced with the question of how to use them If only one trait was analysed there is no problem all animals can be compared on the basis of that single number However when a number of traits have EBV s the breeder has to weigh the importance of each trait to his breeding objective when deciding the overall genetic merit of an animal for his breeding objective PIGBLUP provides INDEX as a tool to assist the breeder in this process Decisions regarding genetic merit must
13. onditions when comparing performance PIGBLUP assigns pigs to management groups in two ways the user defined management group and user recorded management group The former is an implicit grouping of animals that is groups are derived using the maximum group size values chosen by the breeder and the maximum number of weeks for a group while the latter is an explicit grouping 26 AGBU Pigblup Clinic August 1993 A major enhancement of PIGBLUP s analytic capability has been made with the ability to define user management groups This feature gives the breeder greater flexibility in defining management groups such that the animals are assigned appropriately User management groups should be created by the breeder to indicate differing environments affecting the performance of animals Animal production carcase and reproductive groups can be so defined It is important that the number of animals belonging to each management group be neither too few nor too many with too few members there is the possibility of poor estimates of the effect of that management group on an animal s performance with too few different sires and dams genetic effects can be confused with environmental ones and too many animals may result in a lost opportunity to sharpen the resolution by tighter grouping of animals PIGBLUP allows the user to vary the minimum and maximum grouping sizes PIGBLUP automatically initiates a new group when the maximum group size 1s
14. or support 3 2 Flagging Animals In past versions of PIGBLUP flagging was used only to maintain animals in the output that had been selected but had yet to have progeny With the addition of the Genetic Audit and Mate Selection modules flagging has been extended to identify potential animals for mating by placing an A in the appropriate column of the Type 1 PIGBLUP record This is meant to include only the current herd as seen in Figure 1 Some recording systems have been flagging either all animals or all animals who went on to become parents Some of the latter animals are dead and no longer available for mating Also having extra animals marked as active can slow down run times and may even cause the limits for numbers of boars and sows to be exceeded The herd recording system needs the facility to flag as active only live animals from the current herd in the PLGBLUP data extract and to be able to update this in a timely fashion Herd history Back Data Current Herd dead animals live animals With progeny No progeny With progeny Selected but Coming off performance no progeny test data only Live animals available for Mate Selection Active Figure 1 Herd data used by PIGBLUP for analysis 24 AGBU Pigblup Clinic August 1993 3 3 Menu System The PIGBLUP menu system serves a number of purposes It provides a mechanism to smoothly and safely integrate the various modules that make up the PIGBLUP suite It remov

View

Contents

Download Pdf Manuals

Related Search

Related Contents