Traxtile User Manual
Contents
1. 0 0001 72 0 0001 70 0 000252 0 000250 0 000165 0 000163 0 000161 Newick tree format There are several formats used to encode tree like structures The Newick format is one such schema that is widely recognized by tree drawing programs available for Python R and other platforms The set of tracks in an experiment can be reported as a collection of Newick format trees as illustrated below However note that the Newick format which was developed to describe phyolgenies in evolutionary biology does not have a standard notation for the merging of branches 99 32 18 99 20 18 81 25 2 99 23 19 79 23 1 99 121 23 99 115 23 76 115 76 99 13 67 99 38 67 32 19 32 99 34 69 60 31 30 29 23 29 99 62 17 99 59 14 82 58 37 85 58 29 99 65 43 56 56 56 99 79 53 99 71 53 46 78 46 99 62 17 99 59 14 82 58 37 99 70 16 99 84 16 83 81 15 99 70 16 99 84 16 83 81 15 26 79 3 99 27 72 23 72 23 99 46 14 99 53 14 85 48 7 99 26 2 99 36 2 97 26 97 99 100 99 99 123 99 99 120 99 99 124 99 99 102 99 99 107 99 99 103 99 99 114 99 99 105 99 99 110 99 99 111 99 95 100 95 99 109 99 99 119 99 Figure 92. ize OONDOFRWN A O Figure 6 Counts A more detailed report lists data for each frame in the series Each row reports for one time point the numbers of objects cells and the cumulative number of splits newly appearing objects losses i e tips and merges from the beginning of the experiment to the indicated frame The rates as calculated from the start of the experiment to each frame are also given Trends in these calculated rates demonstrate the extent to which estimates have reached stable values during the course of the experiment If they are still changing significantly at the end of the time course a longer period of observation might be helpful This is illustrated in the graph below which was made by pasting the Counts report into Excel Note that the estimate of cell division rate is still rather unstable suggesting that a larger data set more objects and or a longer observation period might be needed SplitRate Ta 60 80 100 120 Time index Figure 7 At the bottom of the Counts report the regression results appear including an estimate of population doubling time based on the regression model If the regression fits a negative exponential curve then the population half life is given instead In the example below direct observation yields an estimate of 4 w of 1 61 x 10 whereas the regression estimates A 1 to be 1 73 x 10 Though one is a positive number and the other negat
3. Save or CSV File Export CSV Required elements for CellProfiler data The following data sources are required to run Traxtile 1 2 3 A comma separated values csv file with object data created by CellProfiler using the ExportToSpreadsheet module A csv file generated by CellProfiler with image data Images to be used for the montage tiles These must be in GIF format and are often saved at some point in the CellProfiler image processing pipeline These files must have the same pixel dimensions as the images used in the dentifyPrimaryObjects module The file name must include a 4 digit number to indicate the time point for the image also called time index or frame index Apart from this 4 digit number the file names must be identical Optionally a separate series of image files to be use for the Image window view These also must be in GIF format have the same pixel dimensions as the images used in IdentifyPrimaryObjects and have names that differ only by the 4 digit time index The following columns are required in the object data file ImageNumber ObjectNumber Location _Center_X Location _Center_Y ParentimageNumber or any column that indicates the image number of an objects parent This is usually the preceding image but it can be further removed if there is a gap at this point in the track The exact name of this column can be specified in Traxtile e ParentObjectNumber or any column tha
4. digit 0 9 In the example above a file named subtracted_2x_s1_t0001 gif was selected The time index is the four digit number after _t so the prefix is subtracted_2x_s1_t and there is no suffix These assignments can be manually edited in the fields provided Pressing OK will trigger reading the files and constructing the data model For large experiments this may take several minutes Then the main tile display will open with the first event s object selected Importing data from other sources The following data sources are required to run Traxtile with Icy e An XLS file output from the Spot Detector plugin converted to CSV format e An XML file created in the TrackManager plugin using the File Save option e Images in GIF format to be used for the montage tiles and optionally the whole image view These are described above The following data sources are required to run Traxtile with Fiji s TrackMate plugin or the ISBI Challenge 2012 format e An XML format file created for example with the Save feature in Trackmate e Images in GIF format to be used for the montage tiles and optionally the whole image view as described above Saving and loading workspace files The File Save menu item allows saving a binary format file containing the data model and configuration information The default extension is txl but any may be used Saved files can be loaded in a
5. section shows the image tiles The center tile shows the event needing review e g the end of a track Tiles that precede or follow show the time steps of the image series and are arranged left to right top to bottom Each tile shows the same cropped region but with different frame numbers The frame numbers are indicated in the lower let corner of each tile The object numbers appear over each object 2 The Information Panel appears below the tiles This shows the identifier of the selected object as well as other information about the selected object including e The lists of parents and children e The extents of the selected cell s linear branch a track segment without splits e The roots and tips connected to the selected cell e Ifthe selected cell is a tip it may be categorized as a Death or Disappearance loss of a cell from the imaging without death e g by migration out of view The object identifier consists of the frame number follwed by a hyphen followed by the object number in that frame Object numbers are assigned by the CellProfiler IdentifyPrimaryObjects module Note that because these are assigned for each image before object tracking object numbers can change from one frame to the next as objects move appear and disappear 3 The Navigation Panel on the bottom provides buttons for moving through the list of reviewed events Loading and Saving data Importing data from CellProfiler The File Import CSV
6. the image file name Importantly the left and right arrow keys on the keyboard will advance through the image series This provides a more dynamic view of object motion than presented by the tile montage Figure 5 Navigating the list of events for review Traxtile compiles a list of sentinel events that are selected for review This list includes New objects that appear after the first frame Tracks that end prior to the last frame Objects that split Objects that merge The bottom of the main window holds buttons for moving forward and back through this list When these are used to advance forward or backward a new set of tiles is displayed centered on the event of interest if possible The object of interest is selected automatically and its track is shown The left and right arrow keys may also be used instead of the buttons The list of events for review is not automatically updated during editing to enable going back to previously reviewed items for a second look However the list can be recompiled by selecting the Analyze Recalculate menu option The list is also recalculated when loading a file from disk or importing CellProfiler data A button marked Go To brings up a dialog box for navigating the list by event number Analysis and reporting A note on modeling population growth A population that is growing or shrinking exponentially follows the equation n t nye wt With definitions n t n
7. the series e Use the dentifyPrimaryObjects module to identify the objects and the TrackObjects module to track them from one frame to the next e Export Image Data and Object Data as spreadsheets i e csv files using the ExportloSpreadsheet module 3 Launch Traxtile An empty workspace will appear 4 Select the File Import CSV menu option and use the dialog box to identify e The CSV file with image data e The CSV file with object data e The columns in the object data file that indicate the Parent Group Parent Object and Frame number Defaults will be selected automatically but the specific names of these columns may vary since they are chosen by CellProfiler using experiment specific metadata e The location of image files to use in the tiles e The location of image files to use for the whole image viewer These may optionally be the same as the tile images e Press OK Traxtile will load and process the data and the tile viewer will fill with images around the first sentinel event object spit merge appearance or disappearance 5 Edit objects and tracks Select a cell by clicking on its number Make or destroy connections to the selected object by click or Alt click Remove the selected object using the Delete key Create new objects using Control click e Move through the list of events using the left and right arrow buttons 6 Create reports using the Analyze menu 7 Save data in binary format File
8. However a connection can be made between the deleted object s parent and child In effect this replaces an object with a gap with no net consequence for the track This is not possible if the deleted object is a root or tip Object creation A new label can be added to the data set using Control click i e holding the Control key and clicking with the mouse in the location for the new label As noted above cells are inferred to exist in gaps so adding cells is often not needed However there may be some cases when a proper timing of a split or death event might require a new object Create or remove connections Editing tracks is the essential purpose of Traxtile In the example below the initial tracking algorithm identified no children for object 64 in frame 27 i e 27 64 so it initially appears as the end of a track Thus it is shown as a track tip and is highlighted with a diamond Figure 2 However this cell actually moved slightly northeast and becomes object 56 in frame 28 because the tracking algorithm did not discover this link 28 56 is erroneously designated as a root having no parents To correct this begin with object 27 64 selected Then holding down 6 OS X or Alt Windows Linux on the keyboard click on 56 in frame 28 This will create the link as shown Figure 3 Now there is a single track that includes 27 64 and 28 56 The data set now has one less tip and one less root Connecti
9. Traxtile User Manual Copyright 2014 The Regents of the University of California All rights reserved Developed by Benjamin S Braun MD PhD UCSF Dept of Pediatrics Installation To run download the Python modules into a single directory and sample data often into a subdirectory but this is not necesary and run traxtile py with the Python 2 7 x interpreter System Requirements Software Python 2 7 x is required earlier versions do not support dictionary comprehension and Python 3 x and above require alternative syntax for some iteration operations Traxtile was developed on the Mac platform Mac OS 10 8 5 but should work on any Python compatible platform with only Python standard libraries Hardware Memory is a limiting resource since the entire list of objects is helpd in memory 2GB is recommended as a minimum but larger amounts may be required for medium sized experiments or larger Multiple processor cores are unlikely to speed execution Hard disk assess time may be rate limiting solid state memory drives can be helpful Quick Start 1 Obtain a series of time lapse images 2 Perform object recognition and initial tracking in CellProfiler e Save at least one processed version of each image in GIF format in a dedicated directory or folder using the Savelmage module The file name must include the frame number in 4 digit string the default is to follow _t e g timelapse_t0123 gif would be image 123 in
10. he subset of those chosen for review i e outside the first and last frames is also given Gains and losses of objects are reported acccording to type both by total number and by the normalized rate per cell per frame To illustrate how rates are calculated consider an example in which 20 objects are followed over 5 frames A total of 100 objects are observed 20 x 5 If one of these splits then the observed rate of splitting is 1 per 100 observations or 0 01 Growth modeling predicts that the exponent in the growth equation should be equal to the net rate of object gains The observed net rate is also reported Finally gaps are summarized by length This information can help guide the choice of gap length parameters when performing object tracking for similar experiments in CellProfiler Note that this information reflects the editing and validation performed by the user and therefore provides a gold standard that can be useful for optimization cells observed starting number ending number 12387 cells x frames 126 124 Ends Roots Tips Total 158 160 Reviewing 32 36 No 13 32 45 Gains splits new cells Total Rate no cell frame 0 00104948736579 0 00258335351578 0 00363284088157 Losses No Rate no cell frame deaths 0 0 0 disappearances 0 0 0 merges 11 0 00088802777105 other 36 0 00290627270526 Total 47 0 00379430047631 Net Rate gains losses 0 000161459594736 Gaps
11. ive both are very close to zero and are not likely to be statistically distinct Frame Cells OONDOABRWN CumCells 126 252 378 504 630 756 882 1008 1134 1259 11396 11521 11645 11769 11893 12016 12139 12263 12387 CumSplits _ OODDDO00 Fitted growth curve n t nO exp a t no exponent a Figure 8 125 007696 1 72979e 05 population doubling time 40071 110188 frames SplitRate 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000882 0 000794 0 001053 0 001042 0 001030 0 001020 0 001009 0 000999 0 001071 0 001060 0 001049 CumNew NewRate ODDO 0O00000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 002808 0 002778 0 002748 0 002719 0 002691 0 002663 0 002636 0 002609 0 002583 CumLoss LossRate OO0OO0OO0OO0OO OO0OOOO 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 002984 0 003038 0 003006 0 002974 0 003027 0 002996 0 002966 0 002936 0 002906 CumMerge MergeRate OO0D90D00000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000794 0 000965 0 000955 0 000945 0 000935 0 000925 0 000915 0 000906 0 000897 0 000888 NetRate 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000000 0 000882 0 000000 0 000088 0 0001 74
12. later session using the File Open menu option Note however that image files must reside in their original location as the saved file refers to this directory Exporting CSV files The current workspace can also be exported as a comma separated values CSV formatted file that can be re imported as an object data file later or used for external analysis For example this file could be used to apply survival analysis to deaths as a function of time with disappearances as censored events Note that this file follows the convention of CellProfiler 2 0 so object merges are not represented explicitly in the CSV file Only one parent is listed for each object For this reason to avoid losing merge events from the data set it is advised to save workspace files in binary format as well Track Editing The following actions are available for editing tracks Object selection An object is selected by clicking on its identifying number The number then is highlighted by color and the tracks associated with that object appear as lines between tiles Object removal The selected object is removed from the database by either pressing the Delete key on the keyboard or pressing the Delete button in the center of the Information Panel When an object is deleted its first child is automatically selected This makes it easy to delete several linked objects in a track There is no undo function so deleting an object is permanent
13. menu option brings up the following dialog box CSV file for Image information Browse directory for data Defau ItOUT_Image csv CSV file for Object information Browse directory for data DefaultOUT_cells csv Field name for Parent Image Number TrackObjects_ParentimageNumber hi Field name for Parent Object Number TrackObjects_ParentObjectNumber Field name for Time Index Metadata_Time Images to use for display Montage tiles Browse directory for gifs prefix extension subtracted_2x_sl_t Whole images Same as montage tile images Browse image extension Cancel The required information is CSV file for Image information The image data file created in CellProfiler using ExoortToSpreadsheet Select a file using the Browse button CSV for Object information The object data file created in CellProfiler using ExportToSpreadsheet Select a file using the Browse button Field name for Parent Image Number The exact name of the column holding the image number for an objects parent corresponding to ParentlmageNumber above Traxtile will attempt to find a column whose title includes the word ParentImageNumber and use this as a default However all the columns in the file are available using the selector Field name for Parent Object Number The exact name of the column holding the object number for an objects
14. ons can be removed the same way Holding down dor Alt and clicking on a connected object will remove the link between it and the selected object There is an alternative way to do this using the Information Panel below the image tiles The parents and children are listed on either side of the slected cell If you press the button marked Unlink Parent then the link to the parent that is highlighted in the Parents list will be removed If no Parent is highlighted then the top one is removed Children 31 21 33 19 31 27 32 19 33 22 Unlink Parent Add Parent Delete Unlink Child Add Child Figure 4 If the Add Parent button is pressed then clicking on an object in the tiled images will be linked as a parent to the current object This is very similar to the click Alt click action There is not a way to manually type in an object identifier to create a link Children can be linked and unlinked the same way Image window It is sometimes helpful to be able to see how objects in a given tile have gotten there especially if they are moving quickly enough that they appear in only a few tiles For assistance the Images window can be opened Window Image menu item or dAlt l This shows an uncropped version of the image containing the selected object The view is centered on the selected object and a highlight box delineates the edges of the tile Scroll bars can be used to pan the image The title of the Image window indicates
15. parent within its own image corresponding to ParentObjectNumber above Traxtile will attempt to find a column whose title includes the word ParentObjectNumber and use this as a default Field name for Time Index The exact name of the column holding the time index for each object also called a frame index as described above for Metadata_time Traxtile will attempt to find a column whose title includes the word Time and use this as a default If image metadata is not stored and if no files were missing during the CellProfiler run then ImageNumber may be selected here Images to use for display Traxtile requires time series images to display in tiles Optionally a second set of images can be used for the Image viewer window though it is common to use the same images for both displays All the images for each purpose must be in a single directory To fill these fields press the Browse button navigate to the proper directory and select any one of the image files in that directory Traxtile will inspect the file name and attempt to separate it into discrete components e Prefix anything preceding the time index e Image the four digit number indicating the time index e Suffix anything coming after the time index before the extension Extension must be gif Except for the image each of these must be identical for all the images The default pattern for the image is _tnnnn where n is a
16. t indicates the parent s object number The exact name of this column can be specified in Traxtile e Metadata_time or any column that holds the time index i e the number indicating the time point in the series This column will only be present if the ExoortloSpreadsheet module in CellProfiler is configured with the option to add image metadata columns to your object data file In addition the time index must be derived from the original image loaded by CellProfiler so that it is included in the image metadata The time index may differ from ImageNumber if there are some missing images in the time lapse series For example if the image file for time point 14 is missing due to corruption or loss then time point 15 will appear with ImageNumber 14 ImageNumber refers to the image number as found by CellProfiler whereas Metadata_time refers to the experimental time point The exact name of this column can be specified in Traxtile If there are no missing image files then ImageNumber may be selected as the time index Any other information in the CellProfiler object data file is not kept in Traxtile but the original CSV file remains unaltered Overview of the user interface Roots Prior Branch Parents Children Next Branch Tips 1 80 1 80 27 64 27 64 Unlink Parent Add Parent Delete Unlink Child Add Child Death Disappear Go To gt Figure 1 The user interface is divided into three sections 1 The main
17. umber of objects at time t Ng number of objects at time zero e base of natural logorithm A birth rate births per cell per time unit u death rate deaths per cell per time unit t time The parameter 2 u is the net growth rate and it can be estimated in Traxtile by two distinct methods First the birth and death rates can be observed directly from the collection of tracks Second the number of objects in each frame can be used to compute an exponential curve by linear regression using the logarithm of obect number Each of these estimates is subject to statistical error so they will not always yield the exact same number Though beyond the scope of this manual these two methods reflect the ability to model population growth using either a deterministic model as produced by the regression or a stochastic Markov birth death model which is informed by the direct observation of events There are three reports available in Traxtile each accessed from the Analyze menu A report is presented as plain text in a spearate window A report can be saved as text files or copied to the system clipboard The report types are Summary This report describes overall statistics for tracks in the series This includes the number of objects present in the first and last images and the total number of ojects identified or inferred in all the images The number of roots and tips is also reported including those in the first and last frames t
Download Pdf Manuals
Related Search