Home

Plato 2.0 User Manual

1. markers Gives the suffix of the marker files for a chromosme in a Beagle fileset See beagle prefix for the naming convention Files compressed with either gzip or bzip2 with an appropriate extension z or gz for gzip and bz for bzip2 will be automatically discovered and decompressed without the need to specify the compression extension beagle files string A comma separated list of Beagle genotype files to load This argument allows the user to load Beagle files that do not adhere to the convention assumed by PLATO and will override any values given by beagle prefix or the suffix arguments This argument must be using in conjunction with marker files and beagle chroms and filenames must be given exactly but automatic decompression will still be performed given the appropriate extensions marker files string A comma separated list of Beagle marker files to load Note that when using this option there must be exactly as many marker files as genotype files and chromosomes given with the beagle files and beagle chroms options As with beagle files filenames must be given exactly but automatic decompression will still be performed given the appropriate extensions beagle chroms string A comma separated list of chromosomes that correspond to the files given by beagle files and beagle markers Note that all 3 arguments must have the same number of files chrom
2. The regression models used by PLATO assume a biallelic marker with a given referent allele In the polyallelic setting all non referent alleles are grouped into a single category which we refer to as the alternate or encoded allele Below are the valid encodings along with a brief description of each e Additive The classic GWAS encoding Homozygous referent is 0 heterozygous is 1 and homozygous alternate is 2 e Dominant Homozygous referent is 0 both heterozygous and homozygous alternate are 1 e Recessive Homozygous referent and heterozygous are 0 while homozygous alternate is 1 Codominant In this encoding each marker uses two variables as a dummy encoding of a categorical variable The Het variable is 1 only when the marker is heterozygous and the Hom variable is 1 only when the marker is homozygous alternate Note that this encoding incurs an additional degree of freedom for each marker used and interactions also incur extra degrees of freedom Weighted This encoding is a hybrid between the traditional encodings and the codominant encoding For each marker the result from a univariate model with appropriate covariates is used to determine an encoding from marker state to the set 0 x 1 where x is chosen such that the model with the encoded allele is identical to the codominant model Then this encoded allele is used in the multivariate models Note that
3. analyses For a handy list of all available commands and options organized by function see the PLATO Quick Reference which lists the commands by function and is cross referenced to their location in the manual Recode PED MAP to Binary PLINK format plato load data file geno output bed file binarygeno Recode Binary PLINK to PED MAP Only a Single Chromosome plato load data bfile bingeno chrom 22 output ped ped geno ped map custom _geno map Perform a Simple Case Control GWAS no covariates Bonferroni corrected p values plato load data bfile bingeno logistic correction Bonferroni PheWAS on Quantitative Outcomes using Markers with MAF between 10 and 40 plato load data bfile bingeno load trait file pheno txt missing 9997 filter maf min 0 1 max 0 4 linear phewas covariates AGE BM EWAS No Genetic Data Quantitative Predictor Variables plato load trait file pheno txt dummy samples extra samples logistic exclude markers use traits outcome PHENO Concordance Checking VCF vs PED MAP w o FID plato load data vcf file data vcf gz concordance file geno no fid GWAS with Dog Data and Permutation Testing plato chroms 38 load data file dog geno logistic permutations 1000 PLATO Quick Reference General Options Special Options h help L list command C help command
4. N GxGxE N N Y Focused GxG Y Y Y Y Focused ExE Y Y Y Y Focused GxGxE Y N dj Error N Y Error
5. cmd v version Global Options f logfile arg plato log chroms 22 extra chroms X Y X Y M MT Display help message List all available PLATO commands Display help for the given command Display version Name of the log file Number of autosomal chromosomes Comma separated list of extra chromosomes dashes indicate aliases Available Commands Data Loading load data load trait Data Transformation recode alleles Data Output output beagle output bed output eigenstrat output ped output tped Filtering filter maf filter marker call filter sample call filter trait missing Analysis batch concordance linear logistic regress auto Loads genetic data into PLATO Loads numeric data into PLATO Set the referent allele for markers Output data in BEAGLE format Output data in binary PLINK bed bim fam format Output data in Eigenstrat format Output data in PLINK ped map format Output data in transposed PLINK tped tfam format Filter markers based on their minor allele frequency Filter markers based on their call rate missing rate Filter samples based on their call rate missing rate Filter traits based on the missing rate Execute PLATO commands from a file Check concordance between two datasets Perform linear regression on the data Perform logistic regression on the data Perform linear or logistic regression based on the phenotype Installation PLATO is packaged
6. excluded the given trait will be excluded from analysis This is especially helpful when used in conjunction with the one sided argument excl traits string Gives a comma separated list of traits to exclude from analysis when generating models This argument may be given multiple times and supersedes incl traits incl markers string Gives a comma separated list of markers to include in the model generation This argument may be given multiple times and is especially helpful when used in conjunction with the one sided argument one sided When used in conjunction with the pairwise argument will instruct PLATO to generate models with one marker or trait if exclude markers was given chosen from the provided list of markers or traits and the other being chosen from all markers or traits not explicitly excluded Model Generation Summary Because the options for generating PLATO models can be somewhat overwhelming the table below lists some of the options available and what the corresponding analysis In the table options are listed by either Y N or which means that the option was given not given or either given or not exclude markers use traits pairwise one sided Analysis N N N N GWAS N Y N N GxE Y Y N N EWAS N N Y N GxG Y Y Y N ExE exclude markers use traits pairwise one sided Analysis N Y Y
7. in the univariate and non interaction case this encoding is identical to the codominant encoding but in the case of interactions incurs fewer degrees of freedom phewas If given PLATO will use all traits not explicity excluded or used as covariates as outcomes in a PheWAS Note that this option is incompatible with the outcome argument correction enum A comma separated list of multiple test correction strategies to use in reporting results from the regression Currently implemented strategies are limited to Bonferroni and FDR output string output txt The filename of the output of the regression The output of the regression is given as a column separated file with one line per model tested sorted by model p value Outcome name if not using the default phenotypic status IDs for the variables of interest For any markers allele and frequency of the encoded allele calculated for the individual model Note that his value may be different for the same marker in different models because of different patterns of missingness The number of missing samples for the model Convergence status if using the logistic command The coefficient standard error and p value for each variable If testing interactions the coefficients will be displayed for both the reduced model main effects only and the full model main effects interaction terms The overall p value of the model s Depending on the arguments use
8. single output for the first two kinds of mismatch and three outputs for the actual call discordance The three discordance outputs are a file of every discordant call and a summary of discordance by sample and by marker Any of the file output can be suppressed by passing an empty string to the appropriate argument The naming convention for the files is lt prefix gt lt suffix gt lt extension gt all of which are customizable below prefix string concordance This option is the prefix that all concordance files will begin with Note that this argument may include directories if the user wishes to place concordance files in a separate directory error string error This argument gives the suffix for the file of every discordant call sample string sample Gives the suffix for the discordance summary by sample marker string marker Gives the suffix for the discordance summary by marker marker mismatch string marker mismatch Gives the suffix for the marker mismatch file sample mismatch string sample mismatch Gives the suffix for the sample mismatch file extension string txt Gives the extension for all concordance output files inc missing If given the concordance command will consider calls that are present in one dataset and missing in the second as discordant By default the concordance checking is performed only on calls that are present in both datasets sep s
9. the parameter passed to this argument This behavior can be overridden on an individual file basis using the ped and map arguments ped Provides a filename for the ped file overriding the file argument map Provides a filename for the map file overriding the map argument output tped This command prints all currently enabled genotype data in transposed PLINK tped tfam format see http pngu mgh harvard edu purcell plink data shtml tr for a complete description of the file format Important Note currently PLATO ignores any genetic distance column 3 of the map file given when loading PLINK genotype files When printing tped files column 4 will ALWAYS be 0 file f string Provides a base filenamefor the transposed PLINK files By default PLATO will output arg tped and arg tfam This behavior can be overridden on an individual file basis using the tped and tfam arguments tped string Provides a filename for the tped file overriding the file argument tfam string Provides a filename for the tfam file overriding the file argument recode alleles This command changes the referent allele for the markers currently loaded by previous commands This is helpful in order to correctly interpret the direction of effect of the coefficients in regression models Note that by changing the referent allele the user will implicitly change the encoded all
10. will print all p values of all permuted models in this file one per line The p values printed will be sorted with the following special p values as flags with the following interpretation e 2 Calculation of the permuted model failed e 3 Could not run permuted model typically too many missing values permu run full By default PLATO will only run enough models in the permutation data to generate an appropriate p value If the user specifies this option PLATO will run all submodels during a permutation This option only has an effect when permuting interaction models and in that case specifying this option will increase the runtime by 33 Model Generation for Regression PLATO can test a variety of models in the regression procedures Typically models are generated by iterating over the dataset and running a separate regression for each model generated The manner in which the user iterates over the dataset will determine the type of analysis being run with possibilities being GWAS EWAS GxE GxG and many others Alternatively the user can individually specify the models to be run which can be helpful in the case of a targeted subset of models The options below are used by the regression options linear and logistic The Model Generation Summary section provides a good summary of the combination of these options and the resultant analysis that can be run using PLATO models string This parameter takes a comma separ
11. Here the option is the option name with the short name given if it exists If the option takes an argument the type of the argument will be provided with the type being one of the following string A character string value Any restrictions on the string filename etc will be specified in the description of the option If the string includes whitespace it should be enclosed with double quotes or it will not be parsed correctly e int An integer value In most cases the integer must be non negative if negative integers are allowed for this parameter it will be explicitly stated in the option description e float A floating point decimal number In most cases only non negative entries make sense but PLATO does not enforce this restriction enum A string value with a very limited set of valid inputs When this argument is seen the description of the option will describle all valid inputs for this parameter When a default value is given this is the value that will be used if the option is not provided on the command line The default values are useful for necessary options such as thresholds and output file names which need values but the user may not want to provide all of the options for compelete customization Examples Below are some examples of analyses that can be run in PLATO and the options that are available to the user These examples are intended as a jumping off point for the user to customize their own
12. NOTE This command is MPI enabled interactions If given include the interaction terms between the variables of interest in the models Note that this argument has no effect if the generated models have only one variable covariates string Provides a comma separated list of traits to include as covariates in each model Covariates given in this argument are premuted along with the phenotypic status when permutation testing is performed for this reason this argument should include phenotypic control variables such as age and gender This argument may be given multiple times const covariates string Provides a comma separated list of traits to include as covariates in the model When using permutation testing these covariates are NOT permutes along with the phenotypic status and should include genotypic control variables such as principal components Note that when not using permutation testing this argument and the covariates argument above are equivalent This argument may be specified multiple times outcome string Provides a comma separated list of traits to use as outcomes for the regression models The generated models will be tested once for every trait given in this argument This argument may be given multiple times If this is not provided PLATO will use the phenotypic status loaded with the data in the load data command encoding enum additive Specifies the encoding of the markers to be used in the models
13. PLATO User Guide Current version PLATO 2 0 Last modified August 2014 Ritchie Lab Center for Systems Genomics Pennsylvania State University University Park PA 16802 URL https ritchielab psu edu Email software ritchielab psu edu Table of Contents O 3 EXA ES a iaussaottuvecncs ETE EE TSE EE TE 4 PLATO Quick REGENCE cccscessssssecsssscssnsoncedescsvecseusosoaseossedseoesonstindesessbecdetedbeeesossosesessesensseesssesessnoddsveses 5 Installation T 6 A a RA E E E cash E A E a a a a E 6 RN 6 A e O 6 Compilation and Instala A Ai E 6 RUN PLATO 0 A Ad 7 Parla A A A A a 7 Spec LOPNA 7 A A tre NRT i PSD SECT NT ME eV OP SR IC R VA nA EA 8 COMMA TIS sce sscceseanadesintcasied A itses seso isses soistes 9 patehan locals A A A ia 9 o O A rte ede indee Muda 9 A Las ha ha chee cence site oh hasta etal St nt ge hat sia o a gh e ee O 10 rai a e a ac 11 tltersample cal aiie tri A DN Nausaanteaddanenens 11 MUSA LOU SO A A AT RIA A sl 11 lingar mnn menaa a n a a tat Dearatinc tile vic tacata a a Ros detent alae a Ta uci te 11 A RA SAA apt E E A AN E a EENAA 14 A RN 19 A SERA A 20 output E a a a Ek aN ne a a iain 21 DP Dd A a 22 A A A E ayes 22 EN 23 A 23 recodo ll A isa 24 A RN 25 Adyanced A TO 25 Pr Filtering DAS A aa cn Nth ie Shere A ni elt sac ON ned cola 25 Permutation Testi aa ia E N aS 21 Model Generation for Represas rt iii ei 28 Overview PLATO the PLatform for the Analysis Translation and Organization of
14. ated list of files containing models to test Each line in the file is assumed to be a whitespace delimited list of variables to include in the model and PLATO assumes that the structure of all models contained in the model files are identical to the first For each variable PLATO attempts to identify by marker ID marker chromosome and position separated by then trait in that order Empty lines and lines beginning with a pound sign are ignored An example of a model file is below This is a comment and ignored All models have 2 markers a trait SAS chr1 342864 BMI EMEZSIOoSOZ Caed A290 LDL Note that order of the variables doesn t matter LDL rs347290 33123606 Also RSIDs are case insensitive and chr is optional for a chrom Rs122483 BMI 4347290 Note that this argument can be given multiple times and if given will override any automated model generation exclude markers If given automatically generated models will not include markers This is useful for an EWAS or an EXE analysis use traits If given will include the traits not listed as covariates in the models to be generated pairwise Generate exhaustive pairwise models incl traits string Gives a comma separated list of traits to include when generating models This argument may be provided multiple times Note that excl traits takes precedence over this argument so if a trait is both included and
15. cessary Parallel Options In PLATO certain commands may use parallelization in order to speed computation times PLATO utilizes a master worker parallelization paradigm where a single computational thread parcels out chunks of data to multiple worker processes who perform the computation and return the result to the master thread which compiles the overall result There are two types of parallelization that PLATO commands may implement threaded shared memory and interprocess distributed memory Threaded commands will typically have the threads option available on the command line and these will require multiple cores on the same machine This option requires less overhead and can substantially speed up computation but it does not scale beyond the number of processors on a single machine To scale beyond a single machine PLATO can use interprocess parallelization using Message Passing Interface MPI Those commands that are capable of MPI parallelization will be specifically noted in the manual all other commands will run in serial when using MPI To use MPI the user must have an MPI compiler installed on their machine during the configuration step PLATO will automatically attempt to enable MPI when it is available Additionally the user must run plato through mpirun as follows mpirun n N plato Above N is the number of processes used by PLATO and the ellipsis indicates that all of the remaining options t
16. d there may be up to 3 distinct p values for the model When interactions are used the overall p value is the likelihood of significance when comparing the full model covariates main effects interactions to the reduced model covariates and main effects only The full and reduced model p values are the likelihood of significance when comparing the respective model to one including covariates only Without interactions there is only a single p value that is calculated comparing the model with variables of interest and covariates to one including only covariates e The corrected p value s of the models The corrected p values are calculated based on the overall model p value see above for the meaning separator string lt TAB gt The string to use for separating columns in the regression output file Show univariate If this argument is provided while examining multivariate models with or without interaction terms PLATO will also perform the appropriate regression using each variable alone and report the results Note that the univariate regressions are not necessarily nested models so the number of missing samples reported for each model may not correspond to the univariate results thresh float 1 Gives a maximum p value for printing models This can be useful for showing only the most significant models in an analysis Note that any multiple test correction strategies given will accurately correct for all tests perf
17. d into memory This can be a helpful check to ensure that the files are complete and correspond to the same IDs logistic This command performs logistic regression on the data previously loaded into PLATO The logistic regression is performed using a Newton Rhapson iteratively reweighted least squares algorithm Many of the options for this command are also applicable to the linear command For the sake of compactness of the manual the options documented below are those that are specific to this command NOTE This command is MPI enabled odds ratio If given PLATO will display the odds ratios for each coefficient rather than the raw coefficient returned by the logistic regression max iterations int 30 Gives the maximum number of least squares steps allowed before declaring nonconvergence of the method Increasing this number will make the regression more robust but may drastically increase the runtime especially in the cases where convergence using a simple logistic regression is impossible output beagle This command will output all currently enabled markers and samples in Beagle file format see http faculty washington edu browning beagle beagle_ 3 3 2 310ct11 pdf for a full description of the file formats Note that phasing and polyallelic data is automatically detected according to the data types loaded into PLATO PLATO will output the genotype and corresponding marker file for each chromosome that contains data T
18. e bed file in individual major mode default is SNP major mode output eigenstrat This command outputs the currently enabled genotype data in Eigenstrat file format see http genepath med harvard edu reich InputFileFormats htm for a full description of the format file f string plato This argument gives the base filename of the Eigenstrat files By default PLATO will output arg geno arg snp and arg indiv for the genotype SNP and individual files respectively This behavior can be 39 66 overridden on an individual file basis using the geno snp and indiv arguments genotype string Provides a filename for the genotype file overriding the file argument Snp string Provides a filename for the SNP file overriding the file argument indiv string Provides a filename for the individual file overriding the file argument output ped This command prints all currently enabled genotype data in traditional PLINK ped map format see http pngu mgh harvard edu purcell plink data shtml for a complete description of the file format Important Note currently PLATO ignores any genetic distance column 3 of the map file given when loading PLINK genotype files When printing map files column 4 will ALWAYS be 0 file f string plato Provides a base for the filenames to print the data to By default PLATO will output arg ped and arg map where arg is
19. ed the marker will NOT be loaded incl sample string This command gives a comma separated list of sample IDs to include for analysis This can be helpful in extracting a small subset from a large dataset Note that if samples are identified by both an IID and an FID the sample ID should include both starting with the FID and they should be whitespace separated For this reason the user may need to enclose this argument in quotes This option may be specified more than once excl sample string This command gives a comma separated list of sample IDs to exclude in any subsequent analysis This can be helpful in excluding problematic samples from analysis such as cryptically related samples As above care must be taken when using sampled identified by both FID and IID and if a sample is both explicitly included and explicitly excluded the sample will NOT be loaded This option may be specified more than once Permutation Testing PLATO offers permutation testing for calculation of the p value of a model in addition to the parametric p value calculation inherent with the regression models The p value for a given model is defined to be the fraction of times any permuted model has a lower p value than the parametrically calculated p value of the unpermuted model Note that the permuted p value should still be subject to multiple test correction and the granularity of the adjusted permuted p value will be determined by the number of per
20. ele file string Provide a two column whitespace separated file of marker IDs and the desired referent allele Markers can also be identified by chromosome and base pair location by using a colon to separate the two Empty lines and lines beginning with a pound sign are ignored Note that if an allele is given but not present for a marker a warning will be issued but the marker will remain unchanged Below is an example of this file format Example recoding allele file TSLIDOL A 36 237630 T chr11 9362856 E Additional lines truncated input map If this argument is given we assume that the file provided above is in extended map format with an additional column providing the referent allele In this case markers are found using the provided ID or the chromosome and base pair position provided in the map file map3 If this argument is provided both input and output map files will not contain the genetic distance column fourth column in a traditional map file auto If given PLATO will set the referent allele as the most common allele present in the dataset In biallelic data this has the effect of making the minor allele the coded allele If the file argument is also given this argument is superseded by those markers found in the recoding file out string Provides a filename to print the marker ID and the referent allele after this step This is particularly useful when automat
21. g lgen arg map and arg fam where arg is the parameter to the lfile argument This can be overridden by the Igen map and fam arguments lgen string Specifies an LGEN file to load Overrides the lfile argument no Sex If given indicates that the PED file or equivalent in the case of binary transposed filesets does not contain gender information typically column 5 no parents If given indicates that the PED file or equivalent does not contain information about parental IDs typically columns 3 and 4 no fid If given indicates that the PED file or equivalent does not contain family ID information typically column 1 Internally the ID used to identify the sample will be the Individual ID repeated no pheno If given indicates that the PED file or equivalent does not contain phenotypic information typically column 6 map3 If given indicates that the MAP file or equivalent does not contain genetic distance information typically column 3 Note that as of PLATO 2 0 genetic distance is ignored even if given map ref If given indicates that the MAP file or equivalent provides the referent allele in the column following the position information typically column 5 Note that in the case of reading a BIM file this is implicitly selected map alt If given indicates that the MAP file or equivalent provides the alternate allele in the colum
22. hat the first column Family ID is missing from the file Internally PLATO will attempt to find the appropriate sample by using the Individual ID repeated as the Sample ID ignore error Allow non numeric values in a trait file and treat those unparseable values as missing Typically PLATO will exit with an error if it encounters a non numeric value this is especially helpful when non numeric values are used to represent missing data such as NA or extra samples If samples are found in the trait file that are not present in the currently loaded simply ignore the line By default PLATO will exit with an error if a sample is present in a trait file but not present in the previously loaded genotype data Note that samples that have previously been marked as disabled are still considered present in the dataset This option is especially helpful when loading phenotype data which may have been collected on more samples that the corresponding genotype data dummy samples Similar to the extra samples option above if this is provided PLATO will create a sample with completely missing genotype data for any samples found in a trait file but not found in the previously loaded genotype data This option is helpful for performing an EWAS or ExE analysis in which the user may not have genotypic data require complete If this option is given PLATO will require an entry in the trait file for all enabled samples currently loade
23. he files will be named according to the pattern lt prefix gt lt chrom gt lt suffix gt where prefix and suffix are user specifiable strings using the arguments below prefix string output This is the prefix to use for all beagle files suffix string bgl This is the suffix to use for all genotype files Note that currently PLATO does not support compressed Beagle file output but this is a feature we hope to add shortly marker suffix string markers This is the suffix to use for all marker files As with the genotype files PLATO does not support output of compressed Beagle format directly incl traits When this argument is given PLATO will include the affection status as well as all currently loaded traits in the genotype files that are output These lines correspond to A and T output lines in the Beagle specification pair Indicates that the data to be output is parent child data which affects the ordering of the samples Note that PLATO will exit with an error if the data loaded does not conform to this structure trio Indicates that the data to be printed is trio data two parents one child Note that PLATO will exit with an error if the data loaded does not conform to this structure missing string 0 Gives the string to use to indicate the presence of a missing allele output bed This command outputs the currently loaded genotype data in binary PLINK bed bim fam forma
24. ically detecting the referent allele and then seeking replication in another dataset as this will ensure that the coded allele is identical in both analyses output map Prints the output above in extended map format This may be useful as part of a pipeline including other tools regress auto This command will perform either a linear or logistic regression based on the phenotype provided to PLATO If there are exactly two non missing phenotypes this command will perform a logistic regression as used in the logistic command If there are more than two unique non missing phenotypes this command will perform a linear regression as in the linear command The output of the regression will include a column indicating the type of regression performed for each model in addition to all of the output columns of the logistic command For linear regression models the Converged column will always be 1 as all linear regressions are not iterative and so they will always converge in one step Note that this command may be dangerous as it can be difficult to compare linear and logistic regression results in a statistically meaningful way linear string This option takes a comma separated list of trait variables for which PLATO should perform linear regression regardless of the number of non missing phenotypes The variables provided to this option and logistic below must be mutually exclusive and given in the outcome o
25. large scale data is analysis software designed to efficiently and easily analyze genome wide genetic data PLATO is designed to be both modular and extensible making it easy to adapt to the ever changing analysis requests of bioinformaticians The core functionality of PLATO is encapsulated in a series of commands specified by the user Each command is processed serially either analyzing loading or printing genotypic data By stringing together a series of commands the user can customize the analysis pipeline to suit their particular needs The convention for calling PLATO on the command line is plato General Options command1 commandl options command2 command2 options This manual will document all of the available commands as well as all of the options available To get started we recommend looking at the examples below as well as the load data or load trait commands which load genetic data and numeric data into PLATO for analysis On the command line options are prefixed with either one or two dashes or for the short or long form of the option All options will have a long form but not all options will have a short form Commands given to PLATO are not prefixed at all and any options seen after the command will be passed to only that command When documenting the options for each command we will use the following convention lt option gt lt short name gt lt argument type gt lt default value gt
26. llowing options affect global settings for how data is stored and the location of files used by all commands Use of the options will not prevent commands from processing but may affect how data is represented and presented to the user logfile f string plato log Gives the name of the log file for this run of PLATO All informational messages warnings and errors printed to the screen will also be printed to the log file so that a user can have a record of each PLATO run chroms int 22 Gives the number of autosomes in the organism for which you are loading genotypic data By default PLATO was designed for human data and uses 22 autosomes extra chroms string X Y XY M MT Provides a comma separated string of non autosomal chromosomes to use in the analysis You may use dashes to indicate aliases for a chromosome and chromosmes are case insensitive For each extra chromosome given PLATO will map it to the next number after the number of autosomes In the default configuration chromosome X is chromosome 23 Y is mapped to chromosome 24 chromosome XY the pseudo autosomal region of the sex chromosomes is mapped to chromosome 25 and chromosome M or MT mitochondrial chromosome is chromosome number 26 Commands In order to do useful work with PLATO a user must enter one or more commands Commands are entered on the command line by beginning an argument without an argument escape or Any arguments beginning wi
27. mutations Thus if a user ran 1 000 permutations the smallest nonzero adjusted permuted p value will be 0 001 The following options control the behavior and output of the permutations run by PLATO permutations int 0 This argument determines the number of permutations per model to run Note that the runtime and memory requirements will increase dramatically with this parameter At a minimum each permutation will take the same amount of time as a single model and will require approximately 12 bytes of memory in lowmem mode For a modest GWAS consisting of 500 000 markers with 1 000 permutations PLATO will need approximately 6 extra GB of RAM and will take 1 000 times as long to execute permu seed int This argument allows for repeatability of results by specifying the initial seed for the random number generator used to initialize the permutations If no seed is given one will be generated at random and reported to the user permu thresh float 0 05 PLATO will print detailed permutation results for all permuted models whose p value falls below this threshold Be aware that increasing this threshold can dramatically increase the storage requirements permu detail fn string permutation detail txt This is the name of the file that PLATO will print the detailed permutation output to To disable this output set this argument equal to the empty string permu pval fn string permutation pval txt PLATO
28. n following the referent allele typically column 6 Note that in the case of reading a BIM file this is implicitly selected control0 If given indicates that the case control phenotypic information from the PED file is encoded as 0 1 where 0 indicates a control and 1 indicates a case By default PLATO assumes a 1 2 encoding where 1 indicates a control and 2 indicates a case with 0 representing a missing value quant If given indicates that the phenotypic information in the PED file is quantitative in nature Note that the case control status of all samples in this case will be unknown beagle prefix string Gives the prefix for loading a set of Beagle files By default PLATO assumes a naming convention of prefix chrom suffix where prefix is the parameter given to this argument and suffix is the parameter given to beagle suffix or marker suffix for the genotype or marker file respectively Additionally chrom is a chromosome string and it is assumed that each genotype file has a corresponding marker file for a given chromosome beagle suffix string bgl Gives the suffix of the genotype files in a Beagle fileset See beagle prefix for the naming convention Files compressed with either gzip or bzip2 with an appropriate extension z or gz for gzip and bz for bzip2 will be automatically discovered and decompressed without the need to specify the compression extension marker suffix string
29. ng allele when loading Beagle genotype files vcf file string Gives the name of the VCF file to be loaded If the VCF file is compressed PLATO will automatically uncompress the data provided that the extension is appropriate i e gz for a b gzipped file or bz for a bzipped file Currently PLATO does not use any associated index files no filter marker By default PLATO will not load any marker that has a FILTER annotation other than PASS or By setting the option PLATO will load all markers without regard to the filter status no filter geno By default PLATO will set any genotype that has a FT annotation other than PASS to missing Including this option will cuase PLATO to load all genotype calls without regard to genotype filter annotation status vcf phased Setting this option tells PLATO that the data in the VCF file is phased Note that PLATO will warn the user if the phasing in the VCF file is inconsistent with the options provided vcf poly Setting this option tells PLATO that the data sin the VCF file is polyallelic If the VCF file is truly polyallelic and this option is not provided PLATO will treat all non referent alleles as the first alternate allele Additionally PLATO will warn if polyallelic markers are detected and this option is not provided load trait This command loads numeric data associated with samples into PLATO for further analysis Note that this command
30. o PLATO remain the same Note that when using MPI N should be greater than 2 as one processor is solely devoted to dividing the work appropriately Additionally by combining threads and mpirun PLATO can use multiple threads on each processor running the work further speeding processing time Special Options The following options are special options that allow PLATO to print help messages If any of the following commands are given no commands provided will be run These arguments are listed below in their order of priority help h Prints some basic help messages This is a good place to start for new users getting used to PLATO list commands L Lists all commands available in the current version of PLATO In the case of a discrepancy between the manual the help documentation in the PLATO program should be considered correct help command C string Prints the help documentation for a given command The help documentation will list ALL available options for a given command While every effort is taken to synchronize the manual to the available PLATO options there may be experimental or undocumented options available that will be shown using this option Note that if no command is given to this option PLATO will print out the help documentation for all available commands version v Prints a brief version string along with a copyright message and a bug reporting email Global Options The fo
31. ormed not just those pritned threads int 1 The number of parallel processes to use in a regression in order to speed runtime Setting this value to 0 disables threading which can be useful in debugging PLATO lowmem If this argument is given PLATO will use temporary files to output some of the results in an attempt to save memory This can be a useful option when testing millions of models but it may increase runtime due to the slow disk speeds when compared to memory load data This function loads genotypic data into PLATO for further processing or during concordance checking Files are loaded from traditional genotype file formats including PLINK and Beagle When parsing options for this command PLATO will load the first set of files that is complete in the following priority order also the order of the commands listed in the help documentation 1 PLINK files ped map 2 Binary PLINK files bed bim fam 3 Transposed PLINK files tped tfam 4 Long format PLINK files lgen map fam 5 Beagle files 6 VCF Files Any options that are given in the section of commands that do not apply to the loaded files will be ignored potentially without warning As an example in the argument combination file myPlink bgl poly the bgl poly argument will have no effect as PLAT will load the PLINK files myPlink ped and myPlink map file string Gives a filename base for reading ped map files By default PLATO
32. osomes trio If given indicates that the data in the Beagle genotype files is trio data with the convention that the data is listed in the order of mother father child Note that although the notation mother and father are used no gender information is assumed If the number of samples given in the genotype file is not a multiple of 3 PLATO will exit with an error This argument is incompatible with the pair argument pair If given indicates that the data in the Beagle genotype files is pair data with the convention that the data is listed in the order of mother child Note that although the notation mother is used no gender information is assumed If the number of samples given in the genotype file is not a multiple of 2 PLATO will exit with an error This argument is incompatible with the trio argument bgl phased If given indicates that the data in the Beagle genotype files is phased Note that biallelic phased data will occupy twice as much memory as biallelic unphased data By default PLATO assumes that Beagle files are biallelic and unphased bgl poly If given indicates that the data in the Beagle genotype file is polyallelic IMPORTANT NOTE Polyallelic data phased or unphased will occupy eight times the memory as biallelic unphased data and four times the memory of phased biallelic data beagle missing string 0 Gives the string to be interpreted as a missi
33. ption if phewas is not provided This option may be specified more than once logistic string This option takes a comma separated list of trait variables for which PLATO should perform logistic regression regardless of the number of phenotypes The variables provided to this option and linear above must be mutually exclusive and given in the outcome option if phewas is not provided This option may be specified more than once Advanced Topics Pre Filtering Data In order to allow analysis on extremely large datasets PLATO provides the option to only load a portion of the dataset provided into memory Note that unlike traditional filters like filter maf or filter sample call if a marker or sample is excluded during the pre filtering stage it cannot be included in any model or later analysis step While these options can reduce the memory footprint of PLATO they will not reduce the time to load data as the entire dataset must be read in order to ensure correct loading of all data chrom string This argument gives a comma separated list of chromosomes to load data for If not given PLATO will load data for all chromosomes bp window srting This argument gives a comma separated list of base pair windows to include in the data loading The windows must be formatted as providing a lower and upper bound inclusive of positions to load data Note that either the upper or lower bound ma
34. should not be used to load numeric data that is categorical in nature for example numbers that represent sites race or ethnicity Examples of numeric data that can be loaded with this command include age lab measurements or quantitative phenotypic measurements such as height weight BMI blood pressure etc file string Gives the filename of the trait file to load A trait file is a whitespace delimited file with each row representing an independent sample The first two columns give the Family ID and the Individual ID for the sample and subsequent columns provide the numeric value Empty lines or lines beginning with a pound sign are ignored and the first line of the file is considered the header line and the values given are the trait IDs that will be available for use in subsequent steps An example of a trait file is below This line is a comment and ignored EILD ILD Age BMI Note the above line which loads 2 traits Age and BMI 11 il 25 DOTA 2 2 34 2 oT S ignore error for how to handle NA 3 3 NA 29 6 S missing for how to handle 9999 4 4 52 9995 Further lines truncated missing string Gives a string value to treat as a missing value This is especially helpful when missing data is represented by a valid number that is nonsensical in the context of the variables being loaded e g 9999 as a value for BMI in the above example no fid Indicates t
35. t see http pngu mgh harvard edu purcell plink binary shtml for a complete description of the format One note in the difference between PLATO and PLINK is that PLATO does not recode the alleles unless explicitly requested in recode alleles Additionally when automatic recoding of alleles is requested the major and referent allele will be printed in column 5 of the bim file and the minor coded allele will be printed in column 6 which is opposite of the behavior of PLINK Despite this difference in behavior both programs will print files that represent the same information and should be 100 concordant Important Note currently PLATO ignores any genetic distance column 3 of the map file given when loading PLINK genotype files When printing bim files column 4 will ALWAYS be 0 file f srting plato This argument gives the base filename of the bed bim bam files By default PLATO will output arg bed arg bim and arg fam files where arg is the given parameter for this argument This behavior can be overridden on an individual file basis using the bed bim or fam arguments bed string Provides an explicit filename for the bed file overriding the file argument bim string Provides an explicit filename for the bim file overriding the file argument fam string Provides an explicit filename for the fam file overriding the file argument individual major Outputs th
36. th the argument escape will be processed according to the most recent command Because some command sequences can get long commands as entered on the command line will be printed in the log file one per line This section will describe all of the available commands in PLATO along with the available options for each command batch The batch command is a way for the user to save a sequence of commands to a file to be reused For those familiar with PLATO 1 x this file is similar to the required batch file file f string This option gives the batch command the name of the file containing commands to process Each batch file must contain a sequence of commands one per line Below is an example of a batch file that performs some basic QC filtering and then runs a logistic regression tilter marker call theesinole 0 99 filter mat min 0 05 mesx 0 4 IOGISLES HOQUIEUE myresulies sde concordance The concordance command will check concordance between the currently loaded dataset and a new dataset to be loaded with this command For the data input options see the load data command There are three distinct types of discordance possible e Sample mismatch A sample is in one dataset but not another e Marker mismatch A marker is present in one dataset but not another e Call discordance The called genotype for a particular sample marker combination does not match across the two datasets This command will produce a
37. tring lt TAB gt Gives the string to use for separating columns in concordance files Useful for automated processing filter maf The filter maf command will set markers as disabled if they do not meet the minor allele frequency thresholds set by the user All markers marked as disabled will be excluded from further analysis including output Important Note This command assumes a biallelic marker and the minor allele frequency is the minimum of the referent allele frequency and one minus the referent allele frequency min float 0 05 Drop all markers with a minor allele frequency below this value default is a MAF of 5 Set to 0 to disable dropping markers based on low minor allele frequency max float 1 0 Drop all markers with a minor allele frequency above this value default is to keep all markers This can be useful in isolating uncommon or rare variants in the dataset filter marker call This command sets makers as disabled according to their missing rate Note that even though a marker has been marked as disabled it can still be explicitly included in models tested by either linear or logistic through the use of model files threshold float 0 99 Drops all markers with a call rate below the given threshold default is 99 or a 1 missing rate filter sample call This command sets samples as disabled according to their missing rate Note that if the data given is pair or trio data this can adversel
38. will load the files arg ped and arg map where arg is the given parameter to the file argument This can be overridden by the ped and map arguments ped string Specifies a PED file to load Overrides the file argument map string Specifies a MAP file to load Overrides the file and Ifile arguments bfile string Gives a filename base for reading bed bim fam files By default PLATO will load the files arg bed arg bim and arg fam where arg is the parameter to the bfile argument This can be overridden by the bed bim and fam arguments bed string Specifies a BED file to load Overrides the bfile argument bim string Specifies a BIM file to load Overrides the bfile argument fam string Specifies a FAM file to load Overrides the bfile and Ifile arguments tfile string Gives a filename base for reading tped tfam files By default PLATO will load the files arg tped and arg tfam where arg is the parameter given to the tfile argument This can be overridden by the tped and tfam arguments tped string Specifies a TPED file to load Overrides the tfile argument tfam string Specifies a TFAM file to load Overrides the tfile argument lfile string Gives a filename base for reading lgen map fam files By default PLATO will load the files ar
39. with the GNU autotools so installation occurs in four steps unpacking configuration compilation and installation Each of those steps will be described below but first the user must ensure that the prerequisites for running PLATO are met Prerequisites The following are prerequisites for building and running PLATO e A modern C compiler e Boost Libraries for C version 1 42 or later http www boost org e Gnu Scientific Library http www gnu org software gsl e optional An MPI enabled compiler and library suite Unpacking PLATO is distributed as a zipped tarball and the command for unpacking the distribution is S tar xvzf plato 2 0 0 tar gz This will unpack the source code into a directory called plato 2 0 0 For all of the following commands we assume that you are in this directory cd plato 2 0 0 Configuration In order to compile PLATO the user must first configure the software This script will attempt to detect all of the prerequisites on the user s system and this is the time for the user to specify system specific options such as the location of the installed program The command is configure The configure script can also take a number of helpful options some of which are detailed below help This option will list all of the available options that can be passed to the configure script prefix path This option tells PLATO to install itself into the given path which is useful if
40. y affect certain downstream analyses such as output beagle threshold float 0 99 Drops all samples with a call rate below the given threshold default is 99 or a 1 missing rate filter trait missing This command drops all traits according to their missing rate Note that if a trait is disabled through the use of this command it is still possible to include it as a covariate in further analyses or to use it explicitly as a variable though the use of a model file in the regression commands threshold float 0 99 Drops all traits with a call rate below the given threshold default is 99 or a 1 missing rate linear This command performs a simple ordinary least squares regression on the data previously loaded into PLATO Most of the options outlined here also apply to the logistic command as well any options that are specific to this command will be noted The strength of PLATO as compared with other available tools is the ability to automatically generate models to perform statistical hypothesis testing By default PLATO will perform a standard GWAS but it can be configured to perform an EWAS GxG GxE or PheWAS analysis depending on the options given See the Model Generation for Regression section for details on how to properly specify the desired analysis type Additionally PLATO can perform permutation testing for details on how to enable this feature see the Permutation Testing section later in the manual
41. y be omitted in which case the bound will be interpreted to mean either 0 or max int respectively Caution should be taken when including base pair windows with multiple chromosomes as the windows will apply to all chromosomes being loaded incl marker string This command gives a comma separated list of marker IDs to include This can be useful when extracting a very small number of SNPs from a large dataset Note that marker IDs can be the traditional RSID or optionally a chrom basepair string This option may be specified more than once excl marker string This command gives a comma separated list of marker IDs to exclude This can be useful when excluding a small number of markers from a dataset Note that if a marker is both explicitly included and explicitly excluded the marker will NOT be loaded This option may be specified more than once incl marker fn string This command gives a comma separated list of files containing a list of marker IDs to include one per line This can be helpful for including a large number of markers for analysis as an example from LD pruning This option may be provided more than once excl marker fn string This command gives a comma separated list of files containing marker IDs to exclude one per line This option can be helpful when excluding a large number of markers as an example when LD pruning Again note that if a marker is explicitly included and explicitly exclud
42. you do not have administrative access to the computer By default the program will be in path bin Note when using this option the path given must be an absolute path and cannot use any shell expansions such as the 66 99 notation enable debug For the advanced users this option will turn off all optimization and turn on debugging symbols which can be helpful in diagnosing a problem with the PLATO software Compilation and Installation Once the configuration of PLATO is complete the user can then compile and install the software To compile PLATO simply use the make command Note that to speed compilation a user may use the j N option to make in order to specify a maximum of N simultaneous parallel processes S make Once compilation is complete PLATO will be available in the current directory Typically the user will need administrative rights to complete the installation step In order to move PLATO to the directory specified by the configuration step type make install Running PLATO After PLATO is installed the user should be able to run PLATO from the command line from any directory If PLATO is not found the installation directory may need to be added to the user s PATH environment variable This section describes the special and global options available to the user These options should be given before any commands on the command line if any special commands are given no commands are ne

Plato 2.0 User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents