When hard-clipping predict when the read is going to be fully hard clipped to the point where only soft/hard-clips are left in the read and preemptively eliminate the read before the SAMRecord mathematics on malformed cigars kills the GATK.
-- GenotypeBuilder now sorts the list of filter strings so that the output is in a consistent order
-- calculateChromosomeCounts removes the AC/AF fields entirely when there are no alt alleles, to be on VCF spec for A defined info field values
-- Fixed bug in VariantDataManager that this validation mode was intended to detect going forward
-- Still no VariantRecalibrationWalkersIntegrationTest for indels with BCF2 but that's because LowQual is missing from test VCF
-- Bugfix for VCFDiffableReader: don't add null filters to object
-- BCF2Codec uses new VCFAlleleClipper to handle clipping / unclipping of alleles
-- AbstractVCFCodec: decodeLoc uses full decode() [still doesn't decode genotypes] to avoid dangerous code duplication. Refactored code that clipped alleles and determined end position into updateBuilderAllelesAndStop method that uses new VCFAlleleClipper. Fixed bug by ensuring the VCF codec always uses the END field in the INFO when it's provided, not just in the case where the there's a biallelic symbolic allele
-- Brand new home for allele clipping / padding routines in VCFAlleleClipper. Actually documented this code, which results in lots of **** negative comments on the code quality. Eric has promised that he and Ami are going to rethink this code from scratch. Fixed many nasty bugs in here, cleaning up unnecessary branches, etc. Added UnitTests in VCFAlleleClipper that actually test the code full. In the process of testing I discovered lots of edge cases that don't work, and I've commented out failing tests or manually skipped them, noting how this tests need to be fixed. Even introduced some minor optimizations
-- VariantContext: validateAllele was broken in the case where there were mixed symbolic and concrete alleles, failing validation for no reason. Fixed.
-- Added computeEndFromAlleles() function to VariantContextUtils and VariantContextBuilder for convenience calculating where the VC really ends given alleles
--
-- refactored allele clipping / padding code into VCFAlleleClipping class, and added much needed docs and TODOs for methods dev guys
-- Added real unit tests for (some) clipping operations in VCFUtilsUnitTest
-- Previous version was reading the size of the encoded genotypes vector for each genotype. This only worked because I never wrote out genotype field values with > 15 elements. Mauricio's killer DiagnoseTargets VCF uncovered the bug. Unfortunately since symbolic allele clipping is still busted those tests are still diabled
-- GenotypeContext getMaxPloidy was returning -1 in the case where there are no genotypes, but the answer should be 0.
-- They now throw an error, as its really unsafe to write out ./. as a special case in the VCFWriter as occurred previously.
-- Added convenience method in VariantContextUtils.addMissingSamples(vc, allSamples) that returns a complete VC where samples are given ./. Genotype objects
-- This allows us to properly pass tests of creating / writing / reading VCFs and BCFs, which previously differed because the VC from the VCF would actually be different from its original VC
-- Updated UG, UGEngine, GenotypeAndValidateWalker, CombineVariants, and VariantsToVCF to manage the master list of samples they are writing out and addMissingSamples via the VCU function
-- Don't use DP for average interval depth but rather AVG_INTERVAL_DP, which is a float now, not an int
-- Don't add PASS filter value to genotypes, as this is actually considered failing filters in the GATK. Genotype filters should be empty for PASSing sites
Updated HSP to use new padding arguments instead of flank intervals file, plus latest QC evals.
IntervalUtils return unmodifiable lists so that utilities don't mutate the collections.
Added a JavaCommandLineFunction.javaGCThreads option to test reducing java's automatic GC thread allocation based on num cpus.
Added comma to list of characters to convert to underscores in GridEngine job names so that GE JSV doesn't choke on the -N values.
JobRunInfo handles the null done times when jobs crash with strange errors.
-- Previously VCF header lines of count type G assumed that the sample would be diploid.
-- Generalized the code to take a VariantContext and return the right result for G count types by calling into the correct numGenotypes in GenotypeLikelihoods class
-- renamed calcNumGenotypes to numGenotypes, which uses a static cache in the class
-- calcNumGenotypes is private, and is used to build the static cache or to compute on the fly for uncached No. allele / ploidy combinations
-- VariantContext calls into getMaxPloidy in GenotypesContext, which caches the max ploidy among samples
-- Added extensive unit tests that compare A and G type values in genotypes
-- Previous bugfix ensures that header fixing is always on in the GATK by default, even after integration tests that failed and when through the VCFDiffableReader. Updating md5s to reflect this.
-- allowMissingVCFHeaders is now part of -U argument. If you want specifically unsafe VCF processing you need -U LENIENT_VCF_PROCESSING. Updated lots of files to use this
-- LENIENT_VCF_PROCESSING disables on the fly VCF header cleanup. This is now implemented via a member variable, not a class variable, which I believe was changing the GATK behavior during integration tests, causing some files to fail that pass when run as a single test because the header reading behavior was changing depending on previous failures.
-- Just completely wrong.
-- BCF2 shadowBCF now checks that the shadow bcf can be written to avoid /dev/null.bcf problem
-- Added samtools ex2.bcf file for decoding to our integrationtests
* field attributesCanBeModified - a null attributes object can't be modified in its current state
* method makeAttributesModifiable() - initialize a null attributes object to empty
-- Added MLEAC and MLEAF format lines to PoolCallerWalker
-- VariantFiltrationWalker now throws an error when JEXL variables cannot be found (XXX < 0.5) but passes through (albeit with a disgusting warning) when a variable is found but its value is a bad type (AF < 0.5) where AF == [0.04,0.00] at multi-allelic variation
-- Allow values to pass assertEquals in VariantContextTestProvider when one file contains X=[null, null] and the other has X missing
-- Update to 2.1.1 from 2.0
-- VariantFiltrationWalker now allows you to run with type unsafe selects, which all default to false when matching. So "AF < 0.5" works even in the presence of multi-allelics now.
--
-- MLAC and MLAF in PoolCaller now use standard MLE_AC and MLE_AF
-- VCFDiffableReader disables onTheFly fixing of VCF header fields so comparisons are easier when headers are changing
-- Flag fields with FLAG_KEY=0 are parsed as though FLAG_KEY were entirely absent in AbstractVCFCodec to fix bug where FLAG_KEY=0 was being translated into FLAG_KEY in output VCF, making a false flag value a true one
-- Fix the GT field value in VariantContextTestProviders so it isn't fixed 1000s of times during testing
-- Keys whose value is null are put into the VariantContext info attributes now
-- Created public static UnifiedGenotyper.getHeaderInfo that loads UG standard header lines, and use this in tools like PoolCaller
-- Created VCFStandardHeaderLines class that keeps standard header lines in the GATK in a single place. Provides convenient methods to add these to a header, as well as functionality to repair standard lines in incoming VCF headers
-- VCF parsers now automatically repair standard VCF header lines when reading the header
-- Updating integration tests to reflect header changes
-- Created private and public testdata directories (public/testdata and private/testdata). Updated tests to use test
-- SelectHeaders now always updates the header to include the contig lines
-- SelectVariants add UG header lines when in regenotype mode
-- Renamed PHRED_GENOTYPE_LIKELIHOODS_KEY to GENOTYPE_PL_KEY
-- Bugfix in BCF2 to handle lists of null elements (can happen in genotype field values from VCFs)
-- Throw error when VCF has unbounded non-flag values that don't have = value bindings
-- By default we no longer allow writing of BCF2 files without contig lines in the header
-- Moved GENOTYPE_KEY vcf header line to VCFConstants. This general migration and cleanup is on Eric's plate now
-- Updated HC to initialize the annotation engine in an order that allows it to write a proper VCF header. Still doesn't work...
-- Updating integration test files. Moved many more files into public/testdata. Updated their headers to all work correctly with new strict VCF header checking.
-- Bugfix for TandemRepeatAnnotation that must be unbounded not A count type as it provides info for the REF as well as each alt
-- No longer add FALSE values to flag values in VCs in VariantAnnotatorEngine. DB = 0 is never seen in the output VCFs now
-- Fixed bug in VCFDiffableReader that didn't differeniate between "." and "PASS" VC filter status
-- Unconditionally add lowQual Filter to UG output VCF files as this is in some cases (EMIT_ALL_SITES) used when the previous check said it wouldn't be
-- VariantsToVCF now properly writes out the GT FORMAT field
-- BCF2 codec explodes when reading symbolic alleles as I literally cannot figure out how to use the allele clipping code. Eric said he and Ami will clean up this whole piece of instructure
-- Fixed bug in BCF2Codec that wasn't setting the phase field correctly. UnitTested now
-- PASS string now added at the end of the BCF2 dictionary after discussion with Heng
-- Fixed bug where I was writing out all field values as BigEndian. Now everything is LittleEndian.
-- VCFHeader detects the case where a count field has size < 0 (some of our files have count = -1) and throws a UserException
-- Cleaned up unused code
-- Fixed bug in BCF2 string encoder that wasn't handling the case of an empty list of strings for encoding
-- Fixed bug where all samples are no called in a VC, in which case we (like the VCFwriter) write out no called diploid genotypes for all samples
-- We always write the number of genotype samples into the BCF2 nSamples header. How we can have a variable number of samples per record isn't clear to me, as we don't have a map from missing samples to header names...
-- Removed old filtersWereAppliedToContext code in VCF as properly handle unfiltered, filtered, and PASS records internally
-- Fastpath function getDisplayBases() in allele that just gives you the raw bytes[] you'd see for an Allele
-- Genotype fields no longer differentiate between unfiltered, filtered, and PASS values. Genotype objects are all PASS implicitly, or explicitly filtered. We only write out the FT values if at least one sample is filtered. Removed interface functions and cleaned up code
-- Refactored padAllele code from createVariantContextWithPaddedAlleles into the function padAllele so that it actually works. In general, **** NEVER COPY CODE **** if you need to share funcitonality make a function, that's why there were invented!
-- Increased the default number of records to read for DiffObjects to 1M
-- The GATK VCFWriter now enforces by default that all INFO, FILTER, and FORMAT fields be properly defined in the header. This helps avoid some of the low-level errors I saw in SelectVariants. This behavior can be disable in the engine with the --allowMissingVCFHeaders argument
-- Fixed broken annotations in TandemRepeat, which were overwriting AD instead of defining RPA
-- Optimizations to VariantEval, removing some obvious low-hanging fruit all in the subsetting of variants by sample
-- SelectVariants header fixes -- Was defining DP for the info field as a FORMAT field, as for AC, AF, and AN original
-- Performance optimizations in BCF2 codec and writer
-- using arrays not lists for intermediate data structures
-- Create once and reuse an array of GenotypeBuilders for the codec, avoiding reallocating this data structure over and over
-- VCFHeader (which needs a complete rewrite, FYI Eric)
-- Warn and fix on the way flag values with counts > 0
-- GenotypeSampleNames are now stored as a List as they are ordered, and the set iteration was slow. Duplicates are detected once at header creation.
-- Explicitly track FILTER fields for efficient lookup in their own hashmap
-- Automatically add PL field when we see a GL field and no PL field
-- Added get and has methods for INFO, FILTER, and FORMAT fields
-- No longer add AC and AF values to the INFO field when there's no ALT allele
-- Memory efficient comparison of VCF and BCF files for shadow BCF testing. Now there's no (memory) constraint on the size of the files we can compare
-- Because of VCF's limited floating point resolution we can only use 1 sig digit for comparing doubles between BCF and VCF
* Sites with more soft clipped bases than regular will force-trigger a variant region
* No more unclipping/reclipping, RR machinery now handles soft clips natively.
* implemented support for base insertion and base deletion quality scores in synthetic and regular reads.
* GATKSAMRecord clone() now creates a fresh object for temporary attributes if one is present.
note: SAMRecords create a shallow copy of the tempAttribute object which was causing multiple reads (that came from the same read) to have their temporary attributes modified by one another inside reduce reads. Beware, if you're not using GATKSAMRecord!
-- Inline encodeString that doesn't go via List<Byte> intermediate
-- Inline encodeString that uses byte[] directly so that we can go from Allele.getBytes() => BCF2
-- Fast paths for Atomic Float and Atomic Integer values avoiding intermediate list creation
-- Final UG integration test update
-- encodeTyped in BCF2Encoder now with specialized versions for int, float, and string, avoiding unnecessary intermediate list creation and dynamic type checking. encodeTypedMissing also includes inline operations now instead of using Collections.emptyList() version. Lots of contracts. User code updated to use specialized versions where possible
-- Misc code refactoring
-- Updated VCF float formating to always include 3 sig digits for values < 1, and 2 for > 1. Updating MD5s accordingly
-- Expanded testing of BCF2Decoder to really use all of the encodeTyped* operations
-- Cleanup a few contracts
-- BCF2FieldManager uses new VCFHeader accessors for specific info and format fields
-- A few simple optimizations
-- VCF header samples stored in String[] in the writer for fast access
-- getCalledChrCount() uses emptySet instead of allocating over and over empty hashset
-- VariantContextWriterStorage now creates a 1MB buffered output writer, which results in 3x performance boost when writing BCF2 files
-- A few editorial comments in VCFHeader
-- Final merge conflicts resolved
-- BCF2Writer now supports case where a sample is present in the header but the sample isn't in the VC, in which case we create an empty sample and encode that
-- Replaced getAttributes with getDP() and not the old style getAttribute, where appropriate
-- Added getAnyAttribute and hasAnyAttribute that actually does the expensive work of seeing if the key is something like GT, AD or another inline datum, and returns it. Very expensive but convenient.
-- Fixed nasty subsetting bug in SelectVariants with excluding samples
-- Generalized VariantsToTable to work with new inline attributes (using getAnyAttribute) as well as GT
-- Bugfix for dropping old style GL field values
-- Added test to VCFWriter to ensure that we have the sample number of samples in the VC as in the header
-- Bugfix for Allele.getBaseString to properly show NO_CALL alleles
-- getGenotypeString in Genotype returns "NA" instead of null for ploidy == 0 genotypes
-- Cleanup some (but not all) VCF3 files. Turns out there are lots so...
-- Refactored gneotype parser from VCFCodec and VCF3Codec into a single shared version in AbstractVCFCodec. Now VCF3 properly handles the new GenotypeBuilder interface
-- Misc. bugfixes in GenotypeBuilder
-- Now only includes leaf nodes in the summary, i.e., summaries of the form "*.*....*.X", which are really the most valuable to see. This calculation can be accomplished in linear time for N differences, rather than the previous O(n^2) algorithm
-- Now computes the max number of elements to read correctly. Counts now the size of the entire element tree, not just the count of the roots, which was painful because the trees vary by orders of magnitude in size.
-- Because of this we can enforce a meaningful, useful value for the max elements in MD5 or 100K, and this works well.
-- Added integration test for new leaf and old pairwise calculations
-- Bugfix for Utils.join(sep, int[]) that was eating the first element of the AD, PL fields
-- BCFFieldEncoder and writers divide up the task of formatting values (atomic or vector, ints, strings, floats, etc) from the task of writing these out at the sites or genotypes level.
-- Allows us to create efficient encoders for specific combinations of header fields, such as int[] encoded values with exactly 3 values
-- Currently only used for INFO fields, but subsequent commit will include optimized genotype field encoder
-- Allowed us to naturally support encoding of lists of strings
-- Bugfixes in VariantContextUtils introduced in genotype -> genotypebuilder conversion
-- Fixes for integration test failures
-- Enabling contig updates
-- WalkerTest now prints out relative paths where possible to make cut/paste/run easier
-- As values in VCs are becoming their native Java types the VCFWriter needs to own proper float formating.
-- Created a smart float formatter in VCFWriter, with unit tests
-- Removed makePrecisionFormatStringFromDenominatorValue and its uses
-- Fix broken contracted
-- Refactored some code from the encoder to utils in BCF2
-- HaplotypeCaller's GenotypingEngine was using old version of subset to context. Replaced with a faster call that I think is correct. Ryan, please confirm.
-- FastGenotypes are the default in the engine. Use --useSlowGenotypes engine argument to return to old representation
-- Cleanup of BCF2Codec. Good error handling. Added contracts and docs.
-- Added a few more contacts and docs to BCF2Decoder
-- Optimized encodePrimitive in BCF2Encoder
-- Removed genotype filter field exceptions
-- Docs and cleanup of BCF2GenotypeFieldDecoders
-- Deleted unused BCF2TestWalker
-- Docs and cleanup of BCF2Types
-- Faster version of decodeInts in VCFCodec
-- BCF2Writer
-- Support for writing a sites only file
-- Lots of TODOs for future optimizations
-- Removed lack of filter field support
-- No longer uses the alleleMap from VCFWriter, which was a Allele -> String, now uses Allele -> Integer which is faster and more natural
-- Lots of docs and contracts
-- Docs for GenotypeBuilder. More filter creation routines (unfiltered, for example)
-- More extensive tests in VariantContextTestProfiler, including variable length strings in genotypes and genotype filters. Better genotype comparisons
-- This file is in integrationtests/md5mismatches.txt, and looks like:
expected observed test
7fd0d0c2d1af3b16378339c181e40611 2339d841d3c3c7233ebba9a6ace895fd test BeagleOutputToVCF
43865f3f0d975ee2c5912b31393842f8 1b9c4734274edd3142a05033e520beac testBeagleChangesSitesToRef
daead9bfab1a5df72c5e3a239366118e 27be14f9fc951c4e714b4540b045c2df testDiffObjects:master=/local/dev/depristo/itest/public/testdata/diffTestMaster.vcf,test=/local/dev/depristo/itest/public/testdata/diffTestTest.vcf,md5=daead9bfab1a5df72c5e3a239366118e
-- Associated cleanup with making md5db an instantiated object, rather than a bunch of static methods
-- decodeIntArray in BCF2 decoder allows us to more efficiently read ints and int[] from stream directly into Genotype object
-- Code cleanup / contracts added were appropriate
-- V2 will have a yet more optimized path...
-- Eliminated the large intermediate map from field name to list of list<Integer> values needed to create genotypes without the GenotypeBuilder. The new code is cleaner and simply fills in an array of GenotypeBuilders as it moves through the column layout in BCF2
-- Now we create once decoders specialized for each GT field (GT, AD, etc) that can be optimized for putting data into the GenotypeBuilder. In a subsequent commit these will actually use lower level BCF2 decoders to create the low-level ints and int[], avoiding the intermediate List<Integer> form
-- Reduced the amount of data further to be computed in the DiffEngine. The DiffEngine algorithm needs to be rethought to be efficient...
-- Builder now provides a depreciated log10pError function to make a new GQ value
-- Genotype is an abstract class, with most of the associated functions implemented here and not in the derived Fast and Slow versions
-- Lots of contracts
-- Bugfixes throughout
-- The way I was handling the contig offset ordering wasn't correct. Now the contigs are always indexed in the order in which their corresponding populate() functions are called, so that the order of the contigs is given by the order in which they are in the file, or in our refDict. It has nothing to do with the contig index itself.
-- SelectVariants no longers prints all samples to the screen if you aren't selecting any explicitly
-- Created a new Genotype interface with a more limited set of operations
-- Old genotype object is now SlowGenotype. New genotype object is FastGenotype. They can be used interchangable
-- There's no way to create Genotypes directly any longer. You have to use GenotypeBuilder just like VariantContextBuilder
-- Modified lots and lots of code to use GenotypeBuilder
-- Added a temporary hidden argument to engine to use FastGenotype by default. Current default is SlowGenotype
-- Lots of bug fixes to BCF2 codec and encoder.
-- Feature additions
-- Now properly handles BCF2 -> BCF2 without decoding or encoding from scratch the BCF2 genotype bytes
-- Cleaned up semantics of subContextFromSamples. There's one function that either rederives or not the alleles from the subsetted genotypes
-- MASSIVE BUGFIX in SelectVariants. The code has been decoding genotypes always, even if you were not subsetting down samples. Fixed!
-- Created new clean FastGenotype and GenotypeBuilder classes with contracts to enforce expected behavior and correctness. Tested utility of this approach by rewritting -- and then commenting out -- a path in BCF2Codec that could use this new code. Much cleaner interface now, but not yet hooked up to anything
-- Disabled SHADOW_BCF generation and generating contigs in the output VCFs automatically to ensure that the current code bases integration tests, before switching the code to new Genotype class
-- Code cleanup. Moved "AD" to VCFConstants under GENOTYPE_ALLELIC_DEPTHS. Uses in code replaced with constant
-- Refactored BCF2Codec into a LazyGenotypesDecoder object that provides on-demand genotype decoding of BCF2 data blocks a la VCFCodec.
-- VCFHeader has getters for sampleNamesInOrder and sampleNameToOffset instead of protected variables directly accessed by vcfcodec
This is in response to a request from Mauricio to make it easier
to use the downsamplers with GATKSAMRecords (as opposed to SAMRecords)
without having to do any cumbersome typecasting. Sadly, Java
language limitations make this sort of solution the best choice.
Thanks to Khalid for his feedback on this issue.
Also:
-added a unit test to verify GATKSAMRecord support with no typecasting required
-added some unit tests for the FractionalDownsampler that Mauricio will/might be using
-moved classes from private to public to better sync up with my local development
branch for engine integration
Moved some stuff in the DiagnoseTargets walker to the more general ThresHolder class
Minor tweaks
FindCoveredIntervals supports Gathering
FindCoveredIntervals outputs an interval list instead of GATKReport
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
* Re-wrote the sliding window approach to allow the variant region not to clip the reads that overlap it.
* Updated consensus to include only reads that were not passed on by the variant region, header counts are updated on the fly to avoid recompute
* Added soft clipped bases to ReduceReads analysis by unclipping high quality soft-clips then re-clipping after reduce reads
* Updated all integration tests
Instead of creating a supposed network temporary directory locally which then fails when remote nodes try to access the non-existant dir, now checking to see if they network directory is available and throwing a SkipException to bypass the test when it cannot be run.
TODO: Throw similar SkipExceptions when fastas are not available. Right now instead of skipping the test or failing fast the REQUIRE_NETWORK_CONNECTION=false means that the errors popup later when the networked fastas aren't found.
- Merged Roger's metrics with Mauricio's optimizations
- Added Stats for DiagnoseTargets
- now has functions to find the median depth, and upper/lower quartile
- the REF_N callable status is implemented
- The walker now runs efficiently
- Diagnose Targets accepts overlapping intervals
- Diagnose Targets now checks for bad mates
- The read mates are checked in a memory efficient manner
- The statistics thresholds have been consolidated and moved outside of the statistics classes and into the walker.
- Fixed some bugs
- Removed rod binding
Added more Unit tests
- Test callable statuses on the locus level
- Test bad mates
- Changed NO_COVERAGE -> COVERAGE_GAPS to avoid confusion
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
-- VCFWriter / codec now passes the same rigorous UnitTest as the BCF2 writer / codec. As part of this we now can only test doubles for equivalence in VCFs to 1e-2 (not exactly impressive)
-- This version of BCF should actually work properly for most files, assuming headers are properly defined.
-- Lots of bug fixes to BCF2 codec
-- Genotype getPhredScaledQual is now an int, returning -1 if there's no QUAL. NOTE THIS SEMANTICS change
-- Equals() method for GenotypeLikelihoods, using PLs.
-- VCFCodec now longer adds empty bindings to missing input field values. NOTE THIS CHANGE
-- VCs can be marked as fully decoded, so that when fullyDecode() is called it returns itself, instead of doing the decoding work. The BCF2 codec now makes VCs marked as fully decoded
-- stringToBytes returns empty list for null or "" string in BCF2Encoder
-- Proper handling of genotype ordering in BCF2 reader / writer
-- Removed the crazy slow noDups and sameSamples tests that were slowing down unit and integration tests totally unnecessarily
-- Many failing MD5s now due to double -> int change in GQ, will update later
-- Added a new parameter to control the maximum number of pairwise differences to generate, which previously could expand to a very large number when there were lots of differences among genotypes, resulting in a n^2 algorithm running with n > 1,000,000
haplotypes were being clipped to the reference window when their unclipped ends went beyond the reference window. The unclipped ends include the hard clipped bases, therefore, if the reference window ended inside the hard clipped bases of a read, the boundaries would be wrong (and the read clipper was throwing an exception).
* updated code to use SoftEnd/SoftStart instead of UnclippedEnd/UnclippedStart where appropriate.
* removed unnecessary code to remove hard clips after processing.
* reorganized the logic to use the assigned read boundaries throughout the code (allowing it to be final).
-- Cut down the size of a few large files in public/testdata that were only used in part
-- Refactor vcf Filename => shadow BCF filename to BCF2Utils. Fix bug in WalkerTest due to the way this was handled previously
-- Fully working version
-- Use -generateShadowBCF to write out foo.bcf as well as foo.vcf anywhere you use -o foo.vcf
-- Moved MedianUnitTest to its proper home in Utils
-- Added reportng to ivy and testng, so build/report/X/html/ is a nicely formatted output for Unit and Integration tests. From this website it's easy to see md5 diffs, etc. This is a vastly better way to manage unit and integration test output
--handle entirely missing GT in a sample in decodeGenotypeAlleles
--Create MAX_ALLELES_IN_GENOTYPES constant in BCF2Utils, and extracted its use inline from the code
-- Generalized genotype writing code to handle ploidy != 2 and variable ploidy among samples
-- Remove special case inline treatment of case where all samples have no GT field values, and moved this into calcVCFGenotypeKeys
-- Removed restriction on getPloidy requiring ploidy > 1. It's logically find to return 0 for a no called sample
-- getMaxPloidy() in VC that does what it says
-- Support for padding / depadding of generic genotype fields
-- fixed final bugs with PL encoding / decoding
-- Ready for testing by other members of the group
-- Current performance numbers aren't so great, but they will improve in the next phase of BCF2 optimizations
-- Fixed a nasty bug in the filter field
-- Not that some (many?) GATK tools won't work with BCF because they internally assume values are Strings not their true types
Read 1500 genotypes file in VCF -> VCF : 11 seconds
Read 1500 genotypes file in VCF -> BCF : 9.5 seconds
VariantEval 1500 genotypes file in VCF : 3 seconds
VariantEval 1500 genotypes file in BCF : 3 seconds
-- Trivial import changes in some walkers
-- SelectVariants has a new hidden mode to fully decode a VCF file
-- DepthPerAlleleBySample (AD) changed to have not UNBOUNDED by A type, which is actually the right type
-- GenotypeLikelihoods now implements List<Double> for convenience. The PL duality here is going to be removed in a subsequent commit
-- BugFixes in BCF2Writer. Proper handling of padding. Bugfix for nFields for a field
-- padAllele function in VariantContextUtils
-- Much better tests for VariantContextTestProvider, including loading parts of dbSNP 135 and the Phase II 1000G call set with genotypes to test encoding / decoding of fields.
-- List<String> is converted inside of the codec to a collapsed string, and exploded in the decoder.
-- Unified the type conversion code in BCFWriter to simply the mapping from VCF type => BCF type and special value recoding
-- Code cleanup and renaming
-- Convenience routine for creating alleles from strings of bases
-- Convenience constructor for VCFFilterHeader line whose description is the same as name
-- VariantContextTestProvider creates all sorts of types of VariantContexts for testing purposes. Can be reused throughtout code for BCF, VCF, etc.
-- Created basic BCF2WriterCodec tests that consumes VariantContextTestProvider contexts, writes them to disk with BCF2 writer, and checks that they come back equals to the original VariantContexts. Actually worked for some complex tests in the first go
-- Added VCFHeader() constructor that makes an empty header, and updated VariantRecalibrator to use it
-- Update build.xml to build vcf.jar with updated paths and bcf2 support.
-- Moved VCF and BCF writers to variantcontext.writers
-- Updated vcf.jar build path
-- Refactored VCFWriter and other code. Now the best (and soon to be only) way to create these files is through a factory method called VariantContextWriterFactory. Renamed the general VCFWriter interface to VariantContextWriter which is implemented by VCFWriter and BCF2Writer.
-- Refactored VCF writers into vcf.writers package
-- Moved BCF2Writer to bcf2.writer
-- Updates to all of the walkers using VCFWriter to reflect new packages
-- A large number of files had their headers cleaned up because of this as well
-- Restructured code to separate the MISSING value in java (currently everywhere a null) from the byte representation on disk (an int).
-- Now handles correctly MISSING qual fields
-- Refactored setting of contigs from VCFWriterStub to VCFUtils. Necessary for proper BCF working
-- Added VCFContigHeaderLine that manages the order for sorting, so we now emit contigs in the proper order.
-- Cleaned up VCFHeader operations
-- BCF now uses the right header files correctly when encoding / decoding contigs
-- Clean up unused tools
-- Refactored header parsing routines to make them more accessible
-- More minor header changes from Intellij
This bug will happen in all adapter/wrapper classes that are passed a resource, and then in their close method they ignore requests to close the wrapped resource, causing a leak when the adapter is the only one left with a reference to the resource.
Ex:
public Wrapper getNewWrapper(File path) {
FileStream myStream = new FileStream(path); // This stream must be eventually closed.
return new Wrapper(myStream);
}
public void close(Wrapper wrapper) {
wrapper.close(); // If wrapper.close() does nothing, NO ONE else has a reference to close myStream.
}
* For some reason, the original implementor decided to use Booleans instead of booleans and didn't always check for null so we'd occasionally get a NPE. Switched over to booleans.
* We'd also generate a NPE if SAMRecord writing specific arguments (e.g. --simplifyBAM) were used while writing to sdout.
The practical differences between version 1.0 and this one (v1.1) are:
* the underlying data structure now uses arrays instead of hashes, which should drastically reduce the memory overhead required to create large tables.
* no more primary keys; you can still create arbitrary IDs to index into rows, but there is no special cased primary key column in the table.
* no more dangerous/ugly table operations supported except to increment a cell's value (if an int) or to concatenate 2 tables.
Integration tests change because table headers are different.
Old classes are still lying around. Will clean those up in a subsequent commit.
* writer mostly implemented
* walkers to convert BCF2 <-> VCF
* almost working for sites-only files; genotypes still need work
* initial performance tests this afternoon will be on sites-only files
From tribble logs:
Binary feature support in tribble
-- Massive refactoring and cleanup
-- Many bug fixes throughout
-- FeatureCodec is now general, with decode etc. taking a PositionBufferedStream
as an argument not a String
-- See ExampleBinaryCodec for an example binary codec
-- AbstractAsciiFeatureCodec provides to its subclass the same String decode,
readHeader functionality before. Old ASCII codecs should inherit from this base
class, and will work without additional modifications
-- Split AsciiLineReader into a position tracking stream
(PositionalBufferedStream). The new AsciiLineReader takes as an argument a
PositionalBufferedStream and provides the readLine() functionality of before.
Could potentially use optimizations (its a TODO in the code)
-- The Positional interface includes some more functionality that's now
necessary to support the more general decoding of binary features
-- FeatureReaders now work using the general FeatureCodec interface, so they can
index binary features
-- Bugfixes to LinearIndexCreator off by 1 error in setting the end block
position
-- Deleted VariantType, since this wasn't used anywhere and it's a particularly
clean why of thinking about the problem
-- Moved DiploidGenotype, which is specific to Gelitext, to the gelitext package
-- TabixReader requires an AsciiFeatureCodec as it's currently only implemented
to handle line oriented records
-- Renamed AsciiFeatureReader to TribbleIndexedFeatureReader now that it handles
Ascii and binary features
-- Removed unused functions here and there as encountered
-- Fixed build.xml to be truly headless
-- FeatureCodec readHeader returns a FeatureCodecHeader obtain that contains a
value and the position in the file where the header ends (not inclusive).
TribbleReaders now skip the header if the position is set, so its no longer
necessary, if one implements the general readHeader(PositionalBufferedStream)
version to see header lines in the decode functions. Necessary for binary
codecs but a nice side benefit for ascii codecs as well
-- Cleaned up the IndexFactory interface so there's a truly general createIndex
function that takes the enumerated index type. Added a writeIndex() function
that writes an index to disk.
-- Vastly expanded the index unit tests and reader tests to really test linear,
interval, and tabix indexed files. Updated test.bed, and created a tabix
version of it as well.
-- Significant BinaryFeaturesTest suite.
-- Some test files have indent changes
-- Other tribble contributors did major refactoring / simplification of tribble, which required some changes to GATK code
-- Integrationtests pass without modification, though some very old index files (callable loci beds) were apparently corrupt and no longer tolerated by the newer tribble codebase
Updated ReadFilter abstract class to implement (via UnsupportedOperationException) the new SamRecordFilter.filterOut().
In IndelRealignerIntegrationTest updates for Picard fixes to SAMRecord.getInferredInsertSize() in svn r1115 & r1124.
- Ran FixMates to create new input BAM since running IR with variable maxReadsInMemory means all reads weren't realigned leading to different outputs.
- Updated md5s to match new expectations after looking at TLEN diff engine output.
* Not working yet, still very much a work-in-progress with lots of placeholders
* Needed to check this in to enable possible collaboration, since it's
going slower than anticipated and the conference deadline looms.
a) Add ability for ErrorModel to be specified by external log-probability vector for testing.
b) For a given depth and ploidy(=2*samples/pool), create artificial high quality pileup testing from AC=0 to AC=ploidy, and test that pool GL's have expected content.Misc. refactorings and cleanups
c) Misc. cleanups and beautification.
a) Utility class called Probability Vector that holds a log-probability vector and has the ability to clip ends that deviate largely from max value.
b) Used this class to hold site error model, since likelihoods of error model away from peak are so far down that it's not worth computing with them and just wastes time.
c) Expand unit tests and add an exhaustive test for ErrorModel class.
d) Corrected major math bug in ErrorModel uncovered by exhaustive test: log(e^x) is NOT x if log's base = 10.
e) Refactored utility functions that created artificial pileups for testing into separate class ArtificialPileupTestProvider. Right now functionality is limited (one artificial contig of 10 bp), can only specify pileups in one position with a given number of matches and mismatches to ref) but functionality will be expanded in future to cover more test cases.
f) Use this utility class for IndelGenotypeLikelihoods unit test and for PoolGenotypeLikelihoods unit test (the latter testing functionality still not done).
g) Linearized implementation of biallelic exact model (very simple approach, similar to diploid exact model, just abort if we're past the max value of AC distribution and below a threshold). Still need to add unit tests for this and to expand to multiallelic model.
h) Update integration test md5's due to minor differences stemming from linearized exact model and better error model math
The GATK -L unmapped is for GenomeLocs with SAMRecord.NO_ALIGNMENT_REFERENCE_NAME, not SAMRecord.getReadUnmappedFlag()
Previously unmapped flag reads in the last bin were being printed while also seeking for the reads without a reference contig.
* fixed queue script plot file names
* updated the ReadGroupCovariate to use the platform unit instead of sample + lane.
* fixed plotting of marginalized reported qualities
* updated BQSR queue script for faster turnaround
* implemented plot generation for scatter/gatherered runs
* adjusted output file names to be cooperative with the queue script
* added the recalibration report file to the argument table in the report
* added ReadCovariates unit test -- guarantees that all the covariates are being generated for every base in the read
* added RecalibrationReport unit test -- guarantees the integrity of the delta tables
* fixed context covariate famous "off by one" error
* reduced maximum quality score to Q50 (following Eric/Ryan's suggestion)
* remove context downsampling in BQSR R script
This test brings together the old and the new BQSR, building a recalibration table using the two separate frameworks and performing the recalibration calculation using the two different frameworks for 10,000+ bases and asserting that the calculations match in every case.
* Refactored CycleCovariate to be a fragment covariate instead of a per read covariate
* Refactored the CycleCovariateUnitTest to test the pairing information
* Updated BQSR Integration tests accordingly
* Made quantization levels parameter not hidden anymore
* Added hidden option to keep intermediate plotting files for debug purposes (they're automatically deleted)
* Added hidden option not to generate the plots automatically (important for scatter/gathering)
The most important reason for this change is that we no longer need to read the entire recal file into memory up front in ApplyRecalibration. For 1000G calling this was prohibitive in terms of memory requirements. Now we go through the rod system and pull in just the records we need at a given position.
As an added bonus, once BCF2 is live we can drastically cut down the sizes of these recal files (which can grow large for whole genome calling).
* removed low quality bases from the recalibration report.
* refactored the Datum (Recal and Accuracy) class structure
* created a new plotting csv table for optimized performance with the R script
* added a datum object that carries the accuracy information (AccuracyDatum) for plotting
* added mean reported quality score to all covariates
* added QualityScore as a covariate for plotting purposes
* added unit test to the key manager to operate with one required covariate and multiple optional covariates
* integrated the plotting into BQSR (automatically generates the pdf with the recalibration tearsheet)
- By porting from jython to java now accessible to Queue via automatic extension generation.
- Better handling for problematic sample names by using PicardAggregationUtils.
GATKReportTable looks up keys using arrays instead of dot-separated strings, which is useful when a sample has a period in the name.
CombineVariants has option to suppress the header with the command line, which is now invoked during VCF gathering.
Added SelectHeaders walker for filtering headers for dbGAP submission.
Generated command line for read filters now correctly prefixes the argument name as --read_filter instead of -read_filter.
Latest WholeGenomePipeline.
Other minor cleanup to utility methods.
-- Not hooked up yet, so the output of VariantEval should be the same as before
-- Implemented a VariantEvalUnitTest that tests the low level strat / eval combinatorics and counting routines
-- Better docs throughout
-- Now properly includes both bi and multi-allelic variants. These are actually counted as well, and emitted as counts and % of sites with multiple alleles
-- Bug fix for gold standard rate
-- HMS no longer tries to grab and throw all exceptions. Exceptions are just thrown directly now.
-- Proper error handling is handled by functions in HMS, which are used by ShardTraverser and TreeReducer
-- Better printing of stack traces in WalkerTest
-- Unfortunately the result of the multi-threaded test is non-deterministic so run the test 10x times to see if the right expection is always thrown
-- Now prints the stack trace and exception message of the caught exception of the wrong type, if this occurs
Adding support for active-region-based annotation for most standard annotations. I need to discuss with Ryan what to do about tests that require offsets into the reads (since I don't have access to the offsets) like e.g. the ReadPosRankSumTest.
IMPORTANT NOTE: this is still very much a dev effort and can only be accessed through private walkers (i.e. the HaplotypeCaller). The interface is in flux and so we are making no attempt at all to make it clean or to merge this with the Locus-Traversal-based annotation system. When we are satisfied that it's working properly and have settled on the proper interface, we will clean it up then.
* Fixed output format to get a valid vcf
* Optimzed the per sample pileup routine O(n^2) => O(n) pileup for samples
* Added support to overlapping intervals
* Removed expand target functionality (for now)
* Removed total depth (pointless metric)
-- SamFileReader.java:525
-- BlockCompressedInputStream:376
These were both instances were we weren't catching and rethrowing picard exceptions as UserExceptions.
- refactored the statistics classes
- concurrent callable statuses by sample are now available.
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
* Added parameter -qq to quantize qualities using a recalibration report
* Added options to quantize using the recalibration report quantization levels, new nLevels and no quantization.
* Updated BQSR scripts to make use of the new parameters
-- Now calculates the number of Indels overlapping gold standard sites, as well as the percent of indels overlapping gold standard sites
-- Removed insertion : deletion ratio for 1 bp event, replaced it with 1 + 2 : 3 bp ratio for insertions and deletions separately. This is based on an old email from Mark Daly:
// - Since 1 & 2 bp insertions and 1 & 2 bp deletions are equally likely to cause a
// downstream frameshift, if we make the simplifying assumptions that 3 bp ins
// and 3bp del (adding/subtracting 1 AA in general) are roughly comparably
// selected against, we should see a consistent 1+2 : 3 bp ratio for insertions
// as for deletions, and certainly would expect consistency between in/dels that
// multiple methods find and in/dels that are unique to one method (since deletions
// are more common and the artifacts differ, it is probably worth looking at the totals,
// overlaps and ratios for insertions and deletions separately in the methods
// comparison and in this case don't even need to make the simplifying in = del functional assumption
-- Added a new VEW argument to bind a gold standard track
-- Added two new stratifications: OneBPIndel and TandemRepeat which do exactly what you imagine they do
-- Deleted random unused functions in IndelUtils
Returns true iff VC is an non-complex indel where every allele represents an expansion or
contraction of a series of identical bases in the reference.
The logic of this function is pretty simple. Take all of the non-null alleles in VC. For
each insertion allele of n bases, check if that allele matches the next n reference bases.
For each deletion allele of n bases, check if this matches the reference bases at n - 2 n,
as it must necessarily match the first n bases. If this test returns true for all
alleles you are a tandem repeat, otherwise you are not. Note that in this context n is the
base differences between the ref and alt alleles
* fixed the loading of the new reduced size reports
* reduced BQSR scala script memory to 2Gb
* removed dcov parameter from BQSR scala script
* fixed estimatedQReported calculation from -log10(pe) to -10*log10(pe).
* updated md5's with the proper PHRED scaled EstimatedQReported
* fixed bug where some keys were using the same recal datum objects
* fixed quantization qual calculations when combining multiple reports
* fixed rounding error with empirical quality reported when combining reports
* fixed combine routine in the gatk reports due to the primary keys being out of order
* added auto-recalibration option to BQSR scala script
* reduced the size of the recalibration report by ~15%
* updated md5's
-- Molten now supports variableName and valueName so you don't have to use variable and value if you don't want to.
-- Cleanup code, reorganize a bit more.
-- Fix for broken integrationtests
*** WAY FASTER ***
-- 3x performance for multiple sample analysis with 1000 samples
-- Analyzing 1MB of the ESP call set (3100 samples) takes 40 secs, compared to several minutes in the previous version
-- According to JProfiler all of the runtime is now spent decoding genotypes, which will only get better when we move to BCF2
-- Remove the TableType system, as this was way too complex. No longer possible to embed what were effectively multiple tables in a single Evaluator. You now have to have 1 table per eval
-- Replaced it with @Molten, which allows an evaluator to provide a single Map from variable -> value for analysis. IndelLengthHistogram is now a @Molten data type. GenotypeConcordance is also.
-- No longer allow Evaluators to use private and protected variables at @DataPoints. You get an error if you do.
-- Simplified entire IO system of VE. Refactored into VariantEvalReportWriter.
-- Commented out GenotypePhasingEvaluator, as it uses the retired TableType
-- Stratifications are all fully typed, so it's easy for GATKReports to format them.
-- Removed old VE work around from GATKReportColumn
-- General code cleanup throughout
-- Updated integration tests
-- Added memory and safety optimizations to StratNode and StratificationManager. Fresh, immutable Hashmaps are allocated for final data structures, so they exactly the correct size and cannot be changed by users.
-- Added ability of a stratification to specify incompatible evaluation. The two strats using this are AC and Sample with VariantSummary, as this computes per-sample averages and so combining these results in an O(n^2) memory requirement. Added integration test to cover incompatible strats and evals
-- Renamed and reorganized infrastructure
-- StratificationManager now a Map from List<Object> -> V. All key functions are implemented. Less commonly used TODO
-- Ready for hookup to VE
-- Created a unit tested tree mapping from a List<String> -> integer (StratificationStates). This class is the key infrastructure necessary to create a complete static mapping from all stratification combinations to an offset in a vector of EvalutionContexts for update in map.
-- Minor code cleanup throughout VE (removing unused headers, for example)
VariantEval is overly abusive of the GATKReport (lack of) spec.
1. It converts numeric values (longs, integers and doubles) to string before sending to the Report, then expects it to decipher that those were actually numbers.
2. Worse, the stratification modules somehow instead of sending the actual values to the report table, sends a string with the value "unknown" and then abuses the GATKReport spec to convert those "unknown" placeholder values with numbers. Then again, it expects the report to know those are numbers, not strings.
Now that the GATKReport HAS specs, VariantEval needs to be overhauled to conform with that. In the meantime, I have added special ad-hoc treatment to these wrong contracts. It works, and the integration tests all passed without changing any MD5's, but right after Mark and Ryan commit their VariantEval refactors, I will step in to change the way it interacts with the GATKReport, so we can clean up the GATKReport.
No wonder, the printing needed to be O(n^2).
* when gathering, be aware that some keys will be missing from some tables.
* when a gatktable has no elements, it should still output the header so we know it had no records
- The Integer column type now accepts byte and shorts
- Updated Unit Tests and added a new testParse() test
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
* restructured the hash tables into one class (RecalibrationReport) that has all the functionality for the different tables and key managers
* optmized empirical qual calculation when merging recalibration reports
* centralized the quality score quantization functionalities
* unified the creating/loading of all the key manager/hash table structures.
* added unit tests for the gatherer (disabled because gatk report needs to be sorted for automated testing)
* added integration tests for BQSR and on-the-fly recalibration
-- Minor refactoring of state key iteration in VEW.map to make the dependencies more clear
-- Long discussion about the performance problems with StateKey, and how to fix it, which I have run out of time to address before ESP meeting.
-- The previous approach (requiring > 5 copies among all reads) is breaking down in many samples (>1000) just from sequencing errors.
-- This breakdown is producing spurious clustered indels (lots of these!) around real common indels
-- The new approach requires >X% of reads in a sample to carry an indel of any type (no allele matching) to be including in the counting towards 5. This actually makes sense in that if you have enough data we expect most reads to have the indel, but the allele might be wrong because of alignment, etc. If you have very few reads, then the threshold is crossed with any indel containing read, and it's counted.
-- As far as I can tell this is the right thing to do in general. We'll make another call set in ESP and see how it works at scale.
-- Added integration tests to ensure that the system is behaving as I expect on the site I developed the code on from ESP
-- StateKey no longer extends TreeMap. It's now a final immutable data structure that caches it's toString and hashcode values. TODO optimizations to entirely remove the TreeMap and just store the HashMap for performance and use the tree for the sorted tostring function.
-- NewEvaluationContext has a method makeStateKey() that contains all of the functionality that once was spread around VEUtils
-- AnalysisModuleScanner uses an annotationCache to speed up the reflections getAnnotations() call when invoked over and over on the same objects. Still expensive to convert each field to a string for the cache, but the only way around that is a complete refactoring of the toTransversalDone of VE
-- VariantEvaluator base class has a cached getSimpleName() function
-- VEUtils: general cleanup due to refactoring of StateKey
-- VEWalker: much better iteration of map data structures. If you need access to iterate over all key/value pairs use the Map.Entry construct with entrySet. This is far better than iterating over the keys and calling get() on each key.
-- Now the only use for update0, calculating the number of processed loci, is centrally tracked in the walker itself not the evaluations.
-- This allows us to avoid calling update0 are every genomic base in 100ks of evaluates when there are a lot of stratifications.
-- No need to modify the integration tests, this optimization doesn't change the result of the calculation
* added empirical quality counts to allow quantization during on-the-fly recalibration to any level
* added number of observations and errors to all tables to enable plotting of all covariates
* restructured BQSR to report recalibrated tables.
* implemented empirical quality calculation to the BQSR stage (instead of on-the-fly recalibration)
* linked quality score quantization to the BQSR stage, outputting a quantization histogram
* included the arguments used in BQSR to the GATK Report
* included all three tables (RG, QUAL and COVARIATES) to the GATK Report with empirical qualities
On-the-fly recalibration with GATK Report
* loads all tables from the GATKReport using existing infrastructure (with minor updates)
* implemented initialiazation of the covariates using BQSR's argument list
* reduced memory usage significantly by loading only the empirical quality and estimated quality reported for each bit set key
* applied quality quantization to the base recalibration
* excluded low quality bases from on-the-fly recalibration for mismatches, insertions or deletions
-- This behavior, which isn't obviously valuable at all, continued to grab and rethrow exceptions in the HMS that, if run without NT, would show up as more meaningful errors. Now HMS simply checks whether the throwable it received on error was a RuntimeException. If so, it is stored and rethrow without wrapping later. If it isn't, only in this case is the exception wrapped in a ReviewedStingException.
-- Added a QC walker ErrorThrowingWalker that will throw a UserException, ReviewedStingException, and NullPointerException from map as specified on the command line
-- Added IT that ensures that all three types are thrown properly (i.e., you catch a NullPointerException when you ask for one to be thrown) with and without threading enabled.
-- I believe this will finally put to rest all of these annoying HMS captures.
-- Use a LinkedHashMap not a TreeMap so iteration is faster.
-- Note that with a lot of stratifications the update0 is taking up a lot of time. For example, with 822 samples and functional class and sample on there are 100K contexts and 30% of the runtime is just in the update0 call
-- Now you always get SNP and indel metrics with VariantEval!
-- Includes Number of SNPs, Number of singleton SNPs, Number of Indels, Number of singleton Indels, Percent of indel sites that are multi-allelic, SNP to indel ratio, Singleton SNP to indel ratio, Indel novelty rate, 1 to 2 bp indel ratio, 1 to 3 bp indel ratio, 2 to 3 bp indel ratio, 1 and 2 to 3 bp indel ratio, Frameshift percent, Insertion to deletion ratio, Insertion to deletion ratio for 1 bp events, Number of indels in protein-coding regions labeled as frameshift, Number of indels in protein-coding regions not labeled as frameshift, Het to hom ratio for SNPs, Het to hom ratio for indels, a Histogram of indel lengths, Number of large (>10 bp) deletions, Number of large (>10 bp) insertions, Ratio of large (>10 bp) insertions to deletions
-- Updated VE integration tests as appropriate
-- Moved a variety of useful formatting routines for ratios, percentages, etc, into VariantEvalator.java so everyone can share. Code updated to use these routines where appropriate
-- Added variantWasSingleton() to VariantEvaluator, which can be used to determine if a site, even after subsetting to specific samples, was a singleton in the original full VCF
-- TableType, which used to be an interface, is now an abstract class, allowing us to implement some generally functionality and avoid duplication.
-- This included creating a getRowName() function that used to be hardcoded as "row" but how can be overridden.
-- #### This allows us implement molten tables, which are vastly easier to use than multi-row data sets. See IndelHistogram class (in later commit) for example of molten VE output
-- No more IndelLengthHistogram (superceded by IndelSummary in subsequent commit)
-- No more SamplePreviousGenotypes or PhaseStats
-- No more MultiallelicAFs
* fixed BadCigarFilter to filter out reads starting/ending in deletion and that have adjacent I/D events.
* added Unit tests for BadCigarFilter
* updated all exceptions in LocusIteratorByState to tell the user that he can instead run with -rf BadCigar
* added the BadCigar filter to ReduceReads and RealignTargetCreator (if your walker blows up with these malformed reads, you may want to add it too)
- Updated the documentation on the code
- Made the table.write() method private and updated necessary files.
- Added a constructor to GATKReport that takes GATKReportTables
- Optimized my code
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
This is important for quick turnaround in the analysis cycle of the new covariates. Also added a dummy unit test that doesn't really test anything (disabled), but helps in debugging.
Pulled out the functionality from Indel Realigner and Table Recalibrator into Utils.setupWriter to make everyone else's life's easier if they want to include the PG tag in their walkers.
Infrastructure:
* Added static interface to all different clipping algorithms of low quality tail clipping
* Added reverse direction pileup element event lookup (indels) to the PileupElement and LocusIteratorByState
* Complete refactor of the KeyManager. Much cleaner implementation that handles keys with no optional covariates (necessary for on-the-fly recalibration)
* EventType is now an independent enum with added capabilities. All functionality is now centralized.
BQSR and RecalibrateBases:
* On-the-fly recalibration is now generic and uses the same bit set structure as BQSR for a reduced memory footprint
* Refactored the object creation to take advantage of the compact key structure
* Replaced nested hash maps with single hash maps indexed by bitsets
* Eliminated low quality tails from the context covariate (using ReadClipper's write N's algorithm).
* Excluded contexts with N's from the output file.
* Fixed cycle covariate for discrete platforms (need to check flow cycle platforms now!)
* Redfined error for indels to look at the previous base in negative strand reads (using new PE functionality)
* Added the covariate ID (for optional covariates) to the output for disambiguation purposes
* Refactored CovariateKeySet -- eventType functionality is now handled by the EventType enum.
* Reduced memory usage of the BQSR script to 4
Tests:
* Refactored BQSRKeyManagerUnitTest to handle the new implementation of the key manager
* Added tests for keys without optional covariates
* Added tests for on-the-fly recalibration (but more tests are necessary)
Infrastructure:
* Generic BitSet implementation with any precision (up to long)
* Two's complement implementation of the bit set handles negative numbers (cycle covariate)
* Memoized implementation of the BitSet utils for better performance.
* All exponents are now calculated with bit shifts, fixing numerical precision issues with the double Math.pow.
* Replace log/sqrt with bitwise logic to get rid of numerical issues
BQSR:
* All covariates output BitSets and have the functionality to decode them back into Object values.
* Covariates are responsible for determining the size of the key they will use (number of bits).
* Generalized KeyManager implementation combines any arbitrary number of covariates into one bitset key with event type
* No more NestedHashMaps. Single key system now fits in one hash to reduce hash table objects overhead
Tests:
* Unit tests added to every method of BitSetUtils
* Unit tests added to the generalized key system infrastructure of BQSRv2 (KeyManager)
* Unit tests added to the cycle and context covariates (will add unit tests to all covariates)
-- TODO for ryan -- there are bugs in ActivityProfile code that I cannot fix right now :-(
-- UnitTesting framework for ActivityProfile -- needs to be expanded
-- Minor helper functions for ActiveRegion to help with unit tests
-- Refactored ART into clearer, simpler procedures. Attempted to merge shared code into utility classes.
-- Added some docs
-- Created a new, testable ActivityProfile that represents as a class the probability of a base being active or inactive
-- Separated band-pass filtering from creation of active regions. Now you can band pass filter a profile to make another profile, and then that is explicitly converted to active regions
-- Misc. utility functions in ActiveRegionWalker such as hasPresetActiveRegions()
-- Many TODOs in ActivityProfile.
-- This report details which intervals are coming in and how many reads they contain
-- Added integration test to verify that the intervals aren't changing, before heading into the ART refactor
GATKReport format changes:
- All non-data header lines are preceeded with a single pound ( #:)
- Every report now has a report header containing the version number and number of tables
- Every table has two lines of table header: The first explains the size of the table and the data types of each column, the second contains the table name and description.
- This new format will allow reports in the future to be gatherable.
- Changed the header format to include an end-of-line string ":;"
Added features:
- Simplified GATK Reports:
The constructor for a simplified GATK Report. Simplified GATK report are designed for reports that do not need the advanced functionality of a full GATK Report.
A simple GATK Report consists of:
- A single table
- No primary key ( it is hidden )
Optional:
- Only untyped columns. As long as the data is an Object, it will be accepted.
- Default column values being empty strings.
Limitations:
- A simple GATK report cannot contain multiple tables.
- It cannot contain typed columns, which prevents arithmetic gathering.
- Added a constructor to generate simplified GATK reports.
- Added a method to easily add data to simple GATK reports.
- Upgraded the input parser take advantage of the new file format (v1).
- Added the GATKReportGatherer, more usability cmoing in next versionof GATK Report. Curently, it can only add rows from one table to another. Added private methods in GATKReport to combine Tables and Reports, It is very conservative and will only gather if the table columns, as well as everything else matches. At the column level, it uses the (redundant) row ids to add new rows. It will throw an exception if it is overwriting data.
- Made some GATKReport methods public, and added more setters and getters.
- Added method that compares formats of two GATKReports, and added an equals method to verify all data inside.
- The gsalib for R now supports reading GATKReport v1 files in addition to legacy formats (v0.*)
- Added a GATKReportDataType enum to give column a certain data type. This must be specified when making a gatherable report. This enum contains several methods including a reverse lookup map.
- Added a data type field in GATKColumn, when a type is not specified, the unknown type is used. Unknown types should not be gathered.
Test changes:
- Updated Unit Tests for GATK Report v1. Added a test for the gatherer. Left one test disabled while we transition from v0 to v1.
- Updated the MD5 hashes in integration tests throughout the GATK.
Other changes:
- Added the gatherer functions to CoverageByRG
- Also added the scatterCount parameter in the Interval Coverage script
- Dropped support for reading in legacy GATKReport formats ( v0.*)
- Updated VariantEvalWalker to work with GATK Report v1, added a format String to all applicable DataPoints.
- Rewrote the read file method for GATK report files.
- Optimized the equals methods within GATKReport. The protected functions should only be called by the GATKReport methods.
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
Now looks like:
<GATK-run-report>
<id>D7D31ULwTSxlAwnEOSmW6Z4PawXwMxEz</id>
<start-time>2012/03/10 20.21.19</start-time>
<end-time>2012/03/10 20.21.19</end-time>
<run-time>0</run-time>
<walker-name>CountReads</walker-name>
<svn-version>1.4-483-g63ecdb2</svn-version>
<total-memory>85000192</total-memory>
<max-memory>129957888</max-memory>
<user-name>depristo</user-name>
<host-name>10.0.1.10</host-name>
<java>Apple Inc.-1.6.0_26</java>
<machine>Mac OS X-x86_64</machine>
<iterations>105</iterations>
</GATK-run-report>
No longer capturing command line or directory information, to minimize people's concerns with phone home and privacy
-- Refactored some duplicated code (FYI, code duplication = root of all evil) into shared functions
-- Added long-missing integrationtests
-- CHRIS/RYAN -- it would be very good to add an integration test covering external VCF files as I believe we rely on this functionality and it's not tested at all
This is a quick-and-dirty patch for the null pointer error Mauricio reported earlier.
Later on we might want to address in a more general way the fact that we validate user intervals
against the reference but not against the merged BAM header produced by the engine at runtime.
This fix is similar, but distinct from the earlier fix to GATKBAMIndex. If we fail to read in
a complete 3-integer bin header from the BAM schedule file that the engine has written, throw a
ReviewedStingException (since this is our problem, not the user's) rather than allowing a
cryptic buffer underflow error to occur.
Note that this change does not fix the underlying problem in the engine, if there is one
(there may be an as-yet-undetected bug in the code that writes the bam schedule). It will
just make it easier for us to identify what's going wrong in the future.
GATKBAMIndex would allow an extremely confusing BufferUnderflowException to be
thrown when a BAM index file was truncated or corrupt. Now, a UserException is
thrown in this situation instructing the user to re-index the BAM.
Added a unit test for this case as well.
-- A cleaner table output (molten). For those interested in seeing how this can be done with GATKReports look here for a nice clean example
-- Integration tests
-- Minor improvements to GATKReportTable with methods to getPrimaryKeys
-- We weren't properly handling the case where a site had both a SNP and indel in both eval and comp. These would naturally pair off as SNP x SNP and INDEL x INDEL in eval, but we'd still invoke update2 with (null, SNP) and (null, INDEL) resulting most conspicously as incorrect false negatives in the validation report.
-- Updating misc. integrationtests, as the counting of comps (in particular for dbSNP) was inflated because of this effect.
Several of the unit tests for the new key authorization feature require
read access to the GATK master private key file. Since this file is only
readable by members of the group gsagit, this makes it hard for people
outside the group to run the test suite.
Now, we skip tests that require the master private key if the private
key exists (since not existing would be a true error) but is not readable
by the user running the test suite
Bamboo, of course, will always be able to run these tests.
-Running the GATK with the -et NO_ET or -et STDOUT options now
requires a key issued by us. Our reasons for doing this, and the
procedure for our users to request keys, are documented here:
http://www.broadinstitute.org/gsa/wiki/index.php/Phone_home
-A GATK user key is an email address plus a cryptographic signature
signed using our private key, all wrapped in a GZIP container.
User keys are validated using the public key we now distribute with
the GATK. Our private key is kept in a secure location.
-Keys are cryptographically secure in that valid keys definitely
came from us and keys cannot be fabricated, however keys are not
"copy-protected" in any way.
-Includes private, standalone utilities to create a new GATK user key
(GenerateGATKUserKey) and to create a new master public/private key
pair (GenerateKeyPair). Usage of these tools will be documented on
the internal wiki shortly.
-Comprehensive unit/integration tests, including tests to ensure the
continued integrity of the GATK master public/private key pair.
-Generation of new user keys and the new unit/integration tests both
require access to the GATK private key, which can only be read by
members of the group "gsagit".
-- Includes paired end status (T/F)
-- Includes count of reads used in calculation
-- Includes simple read type (2x76 for example)
-- Better handling of insert size, read length when there's no data, or the data isn't paired end by emitting NA not 0
-- ReadGroupProperties: Emits a GATKReport containing read group, sample, library, platform, center, median insert size and median read length for each read group in every BAM file.
-- Median tool that collects up to a given maximum number of elements and returns the median of the elements.
-- Unit and integration tests for everything.
-- Making name of TestProvider protected so subclasses and override name more easily
* All contexts with 'N' bases are now collapsed as uninformative
* Context size is now represented internally as a BitSet but output as a dna string
* Temporarily disabled sorted outputs because of null objects
* Turns DNA sequences (for context covariates) into bit sets for maximum compression
* Allows variable context size representation guaranteeing uniqueness.
* Works with long precision, so it is limited to a context size of 31 bases (can be extended with BigNumber precision if necessary).
* Unit Tests added
-- As these represent the bulk of the StingExceptions coming from BAMSchedule and are caused by simple problems like the user providing bad input tmp directories, etc.
-- DoC now by default ignores bases with reference Ns, so these are not included in the coverage calculations at any stage.
-- Added option --includeRefNSites that will include them in the calculation
-- Added integration tests that ensures the per base tables (and so all subsequent calculations) work with and without reference N bases included
-- Reorganized command line options, tagging advanced options with @Advanced
* The tailSet generated every time we flush the reads stash is still being affected by subsequent clears because it is just a pointer to the parent element in the original TreeSet. This is dangerous, and there is a weird condition where the clear will affects it.
* Fix by creating a new set, given the tailSet instead of trying to do magic with just the pointer.
When aggregating raw BAM file spans into shards, the IntervalSharder tries to combine
file spans when it can. Unfortunately, the method that combines two BAM file
spans was seriously flawed, and would produce a truncated union if the file spans
overlapped in certain ways. This could cause entire regions of the BAM file containing
reads within the requested intervals to be dropped.
Modified GATKBAMFileSpan.union() to correct this problem, and added unit tests
to verify that the correct union is produced regardless of how the file spans
happen to overlap.
Thanks to Khalid, who did at least as much work on this bug as I did.
so Ryan can work on the recalibration on the fly without breaking the build. Supposedly all the secret sauce is in the BQSR walker, which sits in private.
* added support to base before deletion in the pileup
* refactored covariates to operate on mismatches, insertions and deletions at the same time
* all code is in private so original BQSR is still working as usual in public
* outputs a molten CSV with mismatches, insertions and deletions, time to play!
* barely tested, passes my very simple tests... haven't tested edge cases.
premature push from my part. Roger is still working on the new format and we need to update the other tools to operate correctly with the new GATKReport.
This reverts commit aea0de314220810c2666055dc75f04f9010436ad.
- Added the GATKReportGatherer
- Added private methods in GATKReport to combine Tables and Reports
- It is very conservative and it will only gather if the table columns, match.
- At the column level it uses the (redundant) row ids to add new rows. It will throw an exception if it is overwriting data.
Added the gatherer functions to CoverageByRG
Also added the scatterCount parameter in the Interval Coverage script
Made some more GATKReport methods public
The UnitTest included shows that the merging methods work
Added a getter for the PrimaryKeyName
Fixed bugs that prevented the gatherer form working
Working GATKReportGatherer
Has only the functional to addLines
The input file parser assumes that the first column is the primary key
Signed-off-by: Mauricio Carneiro <carneiro@broadinstitute.org>
* Adding the context covariate standard in both modes (including old CountCovariates) with parameters
* Updating all covariates and modules to use GATKSAMRecord throughout the code.
* BQSR now processes indels in the pileup (but doesn't do anything with them yet)
* calculates and interprets the coverage of a given interval track
* allows to expand intervals by specified number of bases
* classifies targets as CALLABLE, LOW_COVERAGE, EXCESSIVE_COVERAGE and POOR_QUALITY.
* outputs text file for now (testing purposes only), soon to be VCF.
* filters are overly aggressive for now.
-- This is a partial fix for the problem with uploading S3 logs reported by Mauricio. There the problem is that the java.io.tmpdir is not accessible (network just hangs). Because of that the s3 upload fails because the underlying system uses tmpdir for caching, etc. As far as I can tell there's no way around this bug -- you cannot overload the java.io.tmpdir programmatically and even if I could what value would we use? The only solution seems to me is to detect that tmpdir is hanging (how?!) and fail with a meaningful error.
per Mark's recommendation to reuse the Indel Realigner tag that made it to the SAM spec. The Alignment end tag is still "OE" as there is no official tag to reuse.
* new unit tests for the alignment shift properties of reduce reads
* moved unit tests from ReadUtils that were actually testing GATKSAMRecord, not any of the ReadUtils to it.
* cleaned up ReadUtilsUnitTest
* Added annotations for reads that had been soft clipped prior to being reduced so that we can later recuperate their original alignments (start and end).
* Tags keep the alignment shifts, not real alignment, for better compression
* Tags are defined in the GATKSAMRecord
* GATKSAMRecord has new functionality to retrieve original alignment start of all reads (trimmed or not) -- getOriginalAlignmentStart() and getOriginalAligmentEnd()
* Updated ReduceReads MD5s accordingly
AllelePair is being mean and not telling me what genotype it sees when it finds a non-diploid genotype, but i suspect it's a no-call (".") rather than a no call ("./.").
Eric reported this bug due to the reduced reads failing with an index out of bounds on what we thought was a deletion, but turned out to be a read starting with insertion.
* Refactored PileupElement to distinguish clearly between deletions and read starting with insertion
* Modified ExtendedEventPileup to correctly distinguish elements with deletion when creating new pileups
* Refactored most of the lazyLoadNextAlignment() function of the LocusIteratorByState for clarity and to create clear separation between what is a pileup with a deletion and what's not one. Got rid of many useless if statements.
* Changed the way LocusIteratorByState creates extended event pileups to differentiate between insertions in the beginning of the read and deletions.
* Every deletion now has an offset (start of the event)
* Fixed bug when LocusITeratorByState found a read starting with insertion that happened to be a reduced read.
* Separated the definitions of deletion/insertion (in the beginning of the read) in all UG annotations (and the annotator engine).
* Pileup depth of coverage for a deleted base will now return the average coverage around the deletion.
* Indel ReadPositionRankSum test now uses the deletion true offset from the read, changed all appropriate md5's
* The extra pileup elements now properly read by the Indel mode of the UG made any subsequent call have a different random number and therefore all RankSum tests have slightly different values (in the 10^-3 range). Updated all appropriate md5s after extremely careful inspection -- Thanks Ryan!
phew!
reporting malformed BAMs was actually misreporting any error that happened in the Picard layer
as a BAM ERROR.
Specifically changing PicardException to report as a ReviewedStingException; we might want to
change it in the future. I'll followup with the Picard team to make sure they really, really
want PicardException to inherit from SAMException.
-- Disturbingly, fixing this bug doesn't actually cause an test failures.
-- Wrote a new QCRefWalker to actually check in detail that the reference bases coming into the RefWalker are all correct when comparing against a clean uncached load of the contig bases directly.
-- However, I cannot run this tool due to some kind of weird BAM error -- sending this on to Matt
-- Missing BAMs were appearing as StingExceptions
-- Missing VCFs were showing up as CommandLineErrors, but it's clearer for them to be CouldNotReadInputFile exceptions
-- Added integration tests to ensure missing BAMs, VCFs, and -L files are properly thrown as CouldNotReadInputFile exceptions
-- Added path to standard b37 BAM to BaseTest
-- Cleaned up code in SAMDataSource, removing my parallel loading code as this just didn't prove to be useful.
This is somewhat of an arbitrary decision, and is negotiable. I could see treating
GT:PL ./.:.
differently from
GT:PL .:0,3,6
but am not sure the worth of doing so.
-- Instead issue a warning when a large (>1MB) record is encountered
-- Optimized ref.getBytes()[i] => (byte)ref.charAt(i), which avoids an implicit O(n) allocation each iteration through computeReverseClipping()
-- samtools can emit alleles where the ref is 42M Ns and this caused the GATK (via tribble) to hang in several places.
-- Tribble was updated so we actually could read the line properly (rev. to 51 here).
-- Still the parsing algorithms in the GATK aren't happy with such a long allele. Instead of optimizing the code around an improper use case I put in a limit of 2^16 bp for any allele, and throw a meaningful exception when encountered.
CalibrateGenotypeLikelihoods supports using an external VCF as input for genotype likelihoods. Currently can be a per-sample VCF, but has un-implemented methods for allowing a read-group VCF to be used.
Removed the old constrained genotyping code from UGE -- the trellis calculated is exactly the same as that done in the MLE AC estimate; so we should just re-use that one.
* using the filter() instead of map() makes for a cleaner walker.
* renaming the unit tests to make more sense with the other unit and integration tests
* if the adaptor boundary is more than MAXIMUM_ADAPTOR_SIZE bases away from the read, then let's not clip anything and consider the fragment to be undetermined for this read pair.
* updated md5's accordingly
QScript accessor to QSettings to specify a default runName and other default function settings.
Because log files are no longer pseudo-random their presense can be used to tell if a job without other file outputs is "done". For now still using the log's .done file in addition to original outputs.
Gathered log files concatenate all log files together into the stdout.
InProcessFunctions now have PrintStreams for stdout and stderr.
Updated ivy to use commons-io 2.1 for copying logs to the stdout PrintStream. Removed snakeyaml.
During graph tracking of outputs the Index files, and now BAM MD5s, are tracked with the gathering of the original file.
In Queue generated wrappers for the GATK the Index and MD5s used for tracking are switched to private scope.
Added more detailed output when running with -l DEBUG.
Simplified graphviz visualization for additional debugging.
Switched usage of the scala class 'List' to the trait 'Seq' (think java.util.ArrayList vs. using the interface java.util.List)
Minor cleanup to build including sending ant gsalib to R's default libloc.
To support this, refactored code that computes consensus alleles. To ease merging of mulitple alt alleles, we create a single vc for each alt alleles and then use VariantContextUtils.simpleMerge to carry out merging, which takes care of handling all corner conditions already. In order to use this, interface to GenotypeLikelihoodsCalculationModel changed to pass in a GenomeLocParser object (why are these objects to hard to handle??).
More testing is required and feature turned off my default.
-- Call sets with indels > 50 bp in length are tagged as CNVs in the tag (following the 1000 Genomes convention) and were unconditionally checking whether the CNV is already known, by looking at the known cnvs file, which is optional. Fixed. Has the annoying side effect that indels > 50bp in size are not counted as indels, and so are substrated from both the novel and known counts for indels. C'est la vie
-- Added integration test to check for this case, using Mauricio's most recent VCF file for NA12878 which has many large indels. Using this more recent and representative file probably a good idea for more future tests in VE and other tools. File is NA12878.HiSeq.WGS.b37_decoy.indel.recalibrated.vcf in Validation_Data
This error was due to the ReadClipper change of contract. Before the read utils would return null if a read was entirely clipped, now it returns an empty (safe) GATKSAMRecord.
Mauricio Carneiro
Eric Banks
Ryan Poplin
Mark DePristo
Guillermo del Angel
Khalid Shakir
Roger Zurawicki
Joel Thibault
Christopher Hartl
David Roazen
Laurent Francioli
Menachem Fromer
Andrey Sivachenko
Lechu
Matt Hanna