User Manual for GIGI v1.06.1
Contents
1. know that your file is in the long format you use long flag e g GIGI example param_longFormat txt long Option 1 The long format Each row of the file contains the genotypes for a marker Consistent with the BEAGLE s genotype data file format this file requires a header line The header contains the following information id person1 person1 person2 person2 person3 person3 the id is the name of the marker The pair of columns for each individual represents the first allele and second allele of the specified individual Unlike BEAGLE s genotype file format GIGI s long file format does not have the IP column because GIGI assumes that every row contains a marker An example of this file looks like id 101 101 102 102 103 103 rs0001 111211 rs0002 120012 rs0003222212 For another example see the dense genotypes long file under the Example folder Note if the long format is used the outputs will also be generated in the long format Option 2 The default wide format Each row of the file contains the genotypes for an individual This file does not contain a header line The first column specifies the name of an individual The subject_ID can be an alphanumeric string In this wide format the name of the marker is not retained Subject_ID allelel_marker1 allele2_marker1 allelel_marker2 allele2_marker2 allelel_marker3 allele2_marker3 An example of this file looks like 101111222 1021202. GIGI uses this file which contains IVs at framework positions Note before version 1 06 GIGI also requires the pedigree meiosis file Unless you are using the IVformat31 flag you won t have to prepare the pedigree meiosis file from the console output of gl_auto Refer to the GIGI_v1_04 Documentation 1f you intend to use the older version Obtaining the files a Framework IVs file In gl_auto s parameter file we must instruct gl_auto to display the output as Meiosis Indicators instead of Founder Genome Labels Please make sure we use the option output meiosis indicators instead of output founder genome labels in the control parameter file of gl_auto Installing GIGI Simply unzip the files navigate to the code directory and type make If make does not work go to the GIGI cpp s directory and install the program by g GIGI cpp o GIGI On Linux GIGI was compiled successfully under Ubuntu 10 04 Mandriva Linux release 2009 0 and RedHat release 5 8 with g version 4 1 2 4 3 2 and 4 5 2 On Mac GIGI was compiled successfully under OS X Yosemite 10 10 1 Running GIGI GIGI accepts a parameter file To run GIGI type GIGI lt parameter file gt lt options gt To run the example file go to the main GIGI program directory and type GIGI example param v 1_06 txt Options The flags are case sensitive Flag Purpose verbose to run GIGI with extended output to the console seed NUM _ to change
3. flag longParam when you run GIGI This file format allows you to specify the exact filename and location of the output files e g GIGI example oldparamFormat txt oldParam The old file format line 1 filename of the pedigree meiosis file line 2 number of pedigrees current max is 1 Just use the value 1 line 3 filename of the dense marker file line 4 IVs file line 5 number of saved realizations in the IVs file line 6 filename of the map positions of the framework markers file CM Haldane map distance line 7 filename of the map positions of the dense markers file CM Haldane map distance line 8 filename of the allele frequencies of dense markers file line 9 random seed number e g 1234 line 10 filename of the output 1 called genotypes line 11 Use the value 2 This is the calling method to use 1 most likely genotype calling 2 threshold based calling line 12 the first threshold of threshold based calling t1 to call both alleles this number should be between 0 5 and lt 1 e g 0 9 line 13 the second threshold of threshold based calling t2 to call 1 of 2 alleles if GIGI can t call both alleles under the threshold specified in line 12 a number that is between the first threshold lt 1 e g 0 9 line 14 filename of output 2 estimated genotype probabilities line 15 filename of output 3 number of IVs consistent with each observed dense markers 14 Appendix B If you inte
4. the default seed NUM should be a number between 1 and 999999 prog NUM to print the progress of imputation at every NUM markers oldParam to run GIGI using the old GIGI v1 00 s format of the parameter file General users should ignore In v1 03 and older this flag was called longParam outD to specify the absolute directory path of where the output files will be created If this flag is missing the output files will be saved to the user s current directory e g home charles GIGI_output no character at the end of the directory name long new in v1 04 to read dense genotype marker file as the long format rows are markers and columns are genotypes of individuals IVformat31 new in v 1 06 To use the IV generated by MORGAN gl _ auto version 3 1 or earlier When enabling this option GIGI also requires users to provide the pedigree meiosis file instead of the MORGAN compatible pedigree file An example of running GIGI with additional options GIGI example param v 1_06 txt verbose outD home charles GIGI_output An example of running GIGI with IV formatted from MORGAN gl_auto version 3 1 GIGI example param_v1_05 oldIVformat txt Vformat3 1 Making the Parameter File The parameter file tells GIGI where to look for the required files and where to save the output files GIGI needs the pedigree meiosis file that user prepares from the output of gl_auto dense marker file IVs file from gl_auto ma
5. to t 1 t1 2 The format of the parameter file used here is different than that of an older format used in GIGI_v1 00 See Appendix A for the older format and how we can continue to use the older format of the parameter file not recommended Notes I suggest using the absolute paths of the filenames instead of relative paths A relative path is relative to the directory containing the executable program The parameter file in the example folder is created using a relative path line 1 user creates this file from the gl_auto s console output In version 1 01 GIGI can only process 1 pedigree at a time line 2 when you run gl_auto you should instruct gl_auto to print Meiosis Indicators instead of Founder Genome Labels line 3 this corresponds to the number of samples that the user actually prints to the meisois indicator file File Formats Examples of these files are provided in the example directory refer to the param txt for the filename of these files a pedigree file line 1 This is the pedigree file used by MORGAN gl_ auto to infer IVs For an example of this file please refer to the ped52 MORGAN ped in the example directory Specifically GIGI only cares that 1 the first word in the first line is input to verify that this is a pedigree file 2 the first subject in the pedigree starts on the fifth line with the first 3 columns being subjectID fatherID momID e g input pedigree size 52 input pe
6. 022 103111212 6 allele frequencies of the dense marker file line 7 Each row contains the allele frequencies of the dense markers The first column is the allele frequency of allele 1 the second column is the allele frequency of allele 2 etc Each row must sum to 1 The allele frequencies file is space delimited allele frequencies of marker 1 allele frequencies of marker 2 allele frequencies of marker 3 eg 10 0 4 0 6 0 2 0 3 0 5 0 8 0 2 Output files Three or Four output files are created in the directory where GIGI is run 1 impute geno The imputed genotype file has the same format as the input dense genotype file 2 impute prob The imputed genotype probabilities file displays the genotype probabilities for each marker For each bi allelic marker three columns are used to store the probability of genotype configuration 1 1 2 1 and 2 2 For each tri allelic marker the probabilities are displayed and space delimited with 1 1 2 1 2 2 3 1 3 2 3 3 For multi allelic markers in general the appropriate number of columns are used to store all the combinations Markers are separated by tabs For the long format e g id person personlperson person2 person2 person2 as is specified in the header of the file id is the name of the marker genotype probabilities of different individuals are separated by a tab forthe number of columns correspond to the Note In the header 3 columns are specif
7. GAN GIGI is developed under the linux environment but we have tested that it runs on Mac OSX 10 10 1 In this documentation we use the following terminology Framework markers are a relatively sparse set of markers that are used to infer IVs on a chromosome of interest Dense markers are markers with missing genotypes on some subjects that we want to impute For example these dense markers may be genotypes obtained from sequence data or from a dense SNP panel and may be typed on fewer and even different subjects in the pedigree See the publication describing GIGI below for more information Citing GIGI Cheung CYK Thompson E A Wijsman E M GIGI An approach to effective imputation of dense genotypes on large pedigrees American Journal of Human Genetics 92 4 504 516 Software URL http faculty washington edu wi sman software shtml What s New in version 1 06 1 We have made changes so that GIGI is now compatible with the new Inheritance Vectors file format generated by MORGAN s gl_auto version 3 2 Starting in v1 06 the default behavior of GIGI is to use this new Inheritance Vectors file format In addition because this new file format no longer requires users to provide GIGI with the meiosis indexes GIGI can now directly use the MORGAN pedigree file In the past version that uses MORGAN s gl_auto version 3 1 or earlier users must create the pedigree meiosis file from manually parsing the console output of gl_auto Th
8. User Manual for GIGI v1 06 1 Author Charles Y K Cheung cykc uw edu Ellen M Wijsman wijsman Ouw edu Department of Biostatistics University of Washington Last Modified on 1 31 2015 Contents Ntro d UCI re a tii 3 A PP EREE E ATE ERE TEE ER AR EEE aiia 3 o URD Seir e a A a A EE 3 What s New in yersi n 1 DO nn ri ar i a r aa N aari 3 Files in GIGI software distribution sssini EEN EAN A EA R E Ras 4 T f rrme IVs sme gl Ud id didas 4 T stallins GIG oeae ie E E E AA 5 Runnin GIG Dss e e E o aana aa er E 5 OPUS ncaa E E A ls tias 5 Making the Parameter e a 6 o A tes ian dana senator PP A E 7 Output TES cc eisai Gece cede eect a ed cea deca aeons ce ue teeta esate dd EES 10 Frequently Asked Questions FAQs ad 12 SPUN A asec o o o II o A 13 Appendix Bi iio 14 A A APP oo 14 DICES idad 14 Acknowledgement AA AA usb aac io receeg 15 Introduction GIGI is a C program to impute genotypes in pedigree data It imputes missing genotypes of dense markers conditional on genotypes available on a subset of relatives in the pedigree and framework genotypes used to infer Inheritance Vectors IVs Our pedigree based genotype imputation approach consists of two steps The first step is to infer IVs at the positions of framework markers using gl_auto which is a MCMC based program that is part of the MORGAN package The second step is to impute dense genotypes by GIGI using the IVs output by gl_auto and pedigree structure file from MOR
9. dense markers are the markers that we want to impute on some individuals in the pedigree The dense marker file contains the genotypes of observed individuals a This file is space delimited b The markers should be sorted by ascending map positions c Alleles are labeled numerically starting from 1 2 in ascending order We use 0 to indicate a missing allele Zf the original genotypes are in alpha numeric users first need to convert the genotypes to an indexed numerical format GIGI supports imputation of multi allelic markers new in v1 04 User have two options in the format of the genotype file 1 the newer long format rows are markers or 2 the more traditional wide format rows are individuals The long format should be used for file with many markers e g gt 10 000 Rationale The wide format is not suitable for analysis with a large number of markers If we use the wide format GIGI must read the entire genotype file into the computer memory If the genotype file contains a large number of markers reading the file requires allocation of large memory and may cause problem in memory allocation On the other hand if we use the long format GIGI can process one marker at a time so GIGI can handle arbitrarily large number of markers Therefore we recommend using the long format if the genotype file contains many markers Note The default assumed genotype file format in GIGI is the traditional wide format To let GIGI
10. digree record names 3 integers 2 input pedigree record trait 1 integer 2 KKKKK 1010010 1020020 201 101 10210 202 101 10220 20100020 For both gl_auto and GIGI the pedigree should be ordered in a way such that the ancestors are specified before the descendants For more information please refer to http www stat washington edu thompson Genepi MORGAN morgan3 1 1 tut html morgan tut_2 html SEC13 b framework IVs file line 2 The Inheritance Vectors file describes the descent pattern of chromosomes at the positions of the framework markers It is the output file that gl_auto generates c map positions of framework markers line 4 The marker map positions of the framework markers file is a text file which contains the map distance in centi Morgans cM based on the Haldane map function Markers must be ordered in ascending order and consistent with the order used in gl_auto Each line contains the position of a marker position of Marker position of Marker2 position of Marker3 position of MarkerN eg 1 0 2 0 3 0 4 0 e map positions of dense markers line 5 Similar to the marker map for framework panel of markers the marker map of dense markers is a text file which contains the map distance in centi Morgans cM based on the Haldane map function Markers must be ordered in ascending order Each line contains the position of a marker eg 0 5 0 7 0 9 1 1 1 15 b dense marker genotype file line 6 The
11. e file is formatted as Subject_ID dosage_marker_1 dosage_marker_2 dosage_marker_3 dosage_marker_4 101 0 5 1 0 875 0 102 1 0 5 0 50 Note allele 1 is not necessarily the rare allele It is the user s responsibility to define what allele 1 is 12 Frequently Asked Questions FAQs 1 How do I choose framework markers Because the framework markers are assumed to be sparse we want to choose framework markers that are informative about which chromosomes are being transmitted at the framework loci First framework markers typed on a large number of subjects tend to be most informative Second framework markers that are multi allelic tend to be more informative than di allelic markers if they are available Third 1f framework markers are SNPs markers with high minor allele frequencies in the sample tend to be more informative Fourth framework markers should be moderately but not too dense eg not denser than 1 marker per 0 3 cM because of concern about MCMC mixing and violation of the assumption of Linkage Equilibrium If the framework markers are multiallelic this spacing should be greater e g not denser than 1 marker per 2 cM for good MCMC mixing 2 What do I have to be careful about the genetic map that I am using The map positions are in map distances based on the Haldane map function instead of Kosambi map function or sequence positions If this map is based on the Kosambi map function the user will need to convert the map posi
12. e for more details 15 There is NO WARRANTY for the program to the extent permitted by applicable law In no event unless required by applicable law will any GIGI copyright holder be liable to you for damages including any general special incidental or consequential damages arising out of the use or inability to use the program including but not limited to loss of data or data being rendered inaccurate or losses sustained by you or third parties or a failure of the program to operate with any other programs Acknowledgement Supported by funding from the National Institutes of Health grants R37GM046255 P01HL030086 and P30AG03136 I would also like to thank Elizabeth Marchani Mohamad Saad Alejandro Nato and the Staistical Genetics community for providing valuable feedback as testers of GIGI
13. erefore to avoid confusion we suggest our users to use the pedigree file instead of the pedigree meiosis file from now on Ver 1 06 1 corrected a bug introduced in ver 1 06 in parsing the input information We also understand the importance of backward compatibility If you intend to use gl_auto s output from the pre v3 2 and the pedigree meiosis file you may do so by using the IVformat31 flag Files in GIGI software distribution GIGI software code and its dependency files the Mersenne random number generator e example folder Inferring IVs using gl_auto The first step to impute genotypes is to use framework markers to infer IVs For this purpose we use gl_auto a program in the MORGAN package that is freely available at http www stat washington edu thompson Genepi MORGAN Morgan shtml To infer IVs in gl_auto we need to supply the required files in MORGAN format 1 Pedigree file 2 Marker file this is a composite file that contains the map positions of framework markers in centiMorgans assuming the Haldane map function allele frequencies of framework markers and genotype data of framework markers 3 Parameter file used to run gl_auto Refer to the documentation of MORGAN for guidance on setting up these files and on running gl_auto Example files used to infer IVs using gl_auto are included under the example gl_auto_example directory Overview From running gl_auto we need to obtain the framework IVs file
14. ied for each individual which has the implicit assumption that each marker is di allelic However GIGI does not want to limit imputation to diallelic marker Instead GIGI will continue to output the estimated probabilities of all genotype configurations in multi allelic markers In this case the number of columns gt 3 for that particular multi allelic marker would not match the header A warning is generated in GIGI to remind users to apply caution when interpreting the results from this file For the wide format e g subject_ID genotype_11_markerl genotype_21_markerl genotype_22_markerl genotype_11_marker2 genotype_21_marker2 genotype_22_marker2 1010 90 10 0 8 00 2 1020 90 10 100 11 3 impute consistentIV The consistency file counts the number of IVs consistent with the observed genotypes of each dense marker This file is used for a sanity check A number that is low compared to the number of IVs used may indicate incompatibility between observed genotypes and the sampled IVs 4 impute dosage new in v1 03 If all dense markers to be imputed are di allelic the dosage file will also be created The dosage summarizes the expected percent of the allele 1 in each genotype dosage of a genotype 1 P genotype is 1 1 0 5 P genotype is 1 2 Markers are separated by space For the long format the file is formatted as id dosage_personl dosage_person2 dosage_person3 idis the name of the marker For the wide format th
15. nd to use the IV generated by MORGAN v3 1 or earlier you should use the Pedigree Meiosis file instead of the pedigree file You have to create the Pedigree Meiosis file manually from the console output of gl_auto Using a text editor we open the console_output txt and fetch the line that begins with name name pa name ma Compnt pat meio mat meio We copy this line and table below We paste this table to another file and save it The file includes the header line and looks like this name name pa name ma Compnt pat meio mat meio 2100_6 0 Oo 1 0 0 2100_21 0 0 1 0 0 2100_25 0 0 1 0 0 2100_29 0 0 1 0 0 2100_31 0 0 1 0 0 2100_39 0 0 1 0 0 2100_907 0 Oo 1 0 0 2100_908 2100_901 2100 902 1 2 1 2100_909 0 0 1 0 06 2100_910 2100901 210092 1 4 3 2100 911 2100_901 2100902 1 6 5 2100_915 2100 907 2100908 1 8 7 until the end of table ChangeLog The ChangeLog describes the recent changes to the program It is located in the root directory of the uncompressed file License GIGI is free software you can redistribute it and or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation either version 3 of the License or at your option any later version This program is distributed in the hope that it will be useful but WITHOUT ANY WARRANTY without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public Licens
16. p of the sparse marker file map of the dense marker file and allele frequencies of the dense marker file It produces the called genotype file genotype probability file and another file that prints the number of IVs that are consistent with the observed dense genotypes for sanity check An example of the parameter file is found in the example directory under example param v1_06 txt In GIGI the parameter file is organized as follows line 1 filename of the pedigree file or the pedigree meiosis file if the IVformat31 is used line 2 framework IVs file line 3 number of sampled realizations in the IV file line 4 filename of the map positions of the framework markers file CM based on Haldane map function line 5 filename of the map positions of the dense markers file CM based on Haldane map function line 6 filename of the dense marker genotype file line 7 filename of the allele frequencies of the dense markers file line 8 Call_method t t2 3 numeric values separated by space eg 2 0 6 0 8 On line 8 we specified up to 3 values 1 Call_method use the value 1 to call the most likely genotype and 2 to use threshold based calling 2 and 3 If threshold based calling is used then the threshold of t 0 8 and t2 0 9 are set as defaults This pair of thresholds can be changed by the user by specifying the 2 and 3 numerical values The thresholds should be between 0 5 and lt 1 If only t is specified t will be set
17. tions to Haldane map function using an appropriate conversion method Also since recombination fractions are relative to each other we strongly encourage the user to generate both the framework map positions and the dense marker positions at the same time 3 Can GIGI process multiple pedigrees No In version 1 01 GIGI can only process 1 pedigree at a time For this reason gl_auto should also be run individually for each pedigree 4 Can I use GIGI to impute genotypes on unrelated individuals No 5 What should I know when I use GIGI to impute genotypes on trios Since there is no way to infer recombination on trios using unphased genotypes gl_auto produces an empty inheritance vectors file From the pedigree meiosis file GIGI recognizes that this inheritance vectors file is empty but will proceed to impute genotypes For trios imputing genotypes using GIGI is equivalent to imputing genotypes conditional on the pedigree structure and minor allele frequencies Information about linkage disequilibrium is potentially useful for genotype imputation on trios Since the current version of GIGI does not leverage information from linkage disequilibrium you may want to use existing linkage disequilibrium based software that supports imputation on trios e g BEAGLE 13 Appendix A GIGI v1 00 used an old format for the parameter file It is not recommended for new users For old users if you want to continue to use this old format use the
Download Pdf Manuals
Related Search