zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Commit Graph

  • 4319ff0610 A python script that will convert pooled expanded summary files (from Jason Flannick's pipeline) into .geli files chartl 2009-10-30 16:39:57 +0000
  • a3da475c88 Documentation and cleanup. hanna 2009-10-30 15:40:28 +0000
  • 2d15891719 Created walkers for alignment, validation. hanna 2009-10-30 15:04:07 +0000
  • 51fffc7f69 Comments for Ryan (which also apply to ReadQualityScoreWalker). ebanks 2009-10-30 14:44:04 +0000
  • ccd7440730 We can actually make this a bit simpler (and faster) ebanks 2009-10-30 04:21:03 +0000
  • 1b6333e4ab Enough people have asked for this that it just needed to get written. One can now split up any number of sets into an N-way Venn (although it doesn't check for discordance in the calls, so you'll still want to use SimpleVenn for 2-way comparisons). Wiki docs are updated. ebanks 2009-10-30 04:08:45 +0000
  • 4bdb5b03bd tell UnifiedGenotyper to return calls at all bases ebanks 2009-10-30 03:10:44 +0000
  • 4ee1d6f733 -Have the calculation models determine whether a call passes the lod/confidence thresholds (as opposed to returning everything and letting the UG decide); this way, walkers which call map() will get only the good calls. -Do the right thing in all models for all-base-mode (for Kiran). ebanks 2009-10-30 02:35:51 +0000
  • 64ac956885 Okay, I caved in: CallsetConcordance now gets possible concordance types by looking at classes that implement ConcordanceType instead of having them hard-coded in. Thanks to Kiran this was pretty easy... ebanks 2009-10-30 00:32:26 +0000
  • 1f0d852a48 Fix bug where alignments with indels would be busted because bwa reverses the read bases to undo a previous read base reverse that doesn't occur in the libbwa codepath. Also fixed some memory management issues. hanna 2009-10-29 21:33:13 +0000
  • ef00ba3ced added IndelGenotyperV2Walker asivache 2009-10-29 20:07:31 +0000
  • e3b4d4cbed Genotyper reimplemented. Does the same thing, at least for now, but internal data structures redesign enables collecting various statistics for indel-containing/reference-matching reads. The statistics are not yet used by the caller itself to make a better judgement w.r.t. the validity of the calls it makes, but they are now printed into the output stream (--verbose). The statistics (for both normal and tumor) include: indel observation count/total coverage, av. number of mismatches per indel-containing and per ref-matching read, av. mapping quality, av. mismatch rate and av. base quality within an NQS windoew around the indel, numbers of indel and ref observations per strand. asivache 2009-10-29 19:09:16 +0000
  • c35a457a09 Delete duplicate jgrapht reference. hanna 2009-10-29 17:38:01 +0000
  • f04b80d7db Fixed epic memory leak. hanna 2009-10-29 16:32:43 +0000
  • 2b96b2e4e7 better multi-sample integration test ebanks 2009-10-29 13:51:51 +0000
  • 1c4ca9d383 -Mark just reminded me: actually force the ref/loc to be immutable -VCF writer should be blind to the score/confidence/lod value - just print the thing out as is ebanks 2009-10-29 13:41:53 +0000
  • 5cdbdd9e5b now that the design is stable, pull the setReference and setLocation methods back out of Genotype and stick them into constructors of implementing classes ebanks 2009-10-29 13:27:37 +0000
  • 3091443dc7 Sweeping changes to the genotype output system, as per several discussions with Matt & Aaron. Some things still need to be changed, but it will entail some more design decisions first (which means I get to bug M&A again tomorrow!). ebanks 2009-10-29 03:46:41 +0000
  • 86573177d1 Reverting rod walkers to use underlying refwalker implementation while we work on ROD2 and reenable the system. Added some serious sparse file parsing to variant eval tests depristo 2009-10-29 01:04:37 +0000
  • c9a3707cfd Initial version of BWA/C bindings. Still lots of squirrels roaming the code. - Some cigar strings aren't right. - Memory leaks. - BWA codebase changes aren't committed to BWA tree. - Aligner interface butchered to support BWA/C-style alignments. hanna 2009-10-28 21:37:49 +0000
  • c4359bc340 Whoops. Forgot the implements. chartl 2009-10-28 19:59:57 +0000
  • 5a3bd50537 adding error log reporting to the GATK, and a stream based output method for the argument collection aaron 2009-10-28 19:56:05 +0000
  • 863d3023d5 IndelCounterWalker -- a new little walker that counts indels over a region (want to see what kind of havoc BWA may be resulting in). Don't know when BasicPileup.indelPileup() was written, but kudos to whoever wrote it. chartl 2009-10-28 19:50:50 +0000
  • 04e9a494e9 removed the GenotypesBacked interface, which is currently unused. Also cleaned up some documentation lines aaron 2009-10-28 18:08:14 +0000
  • e3b9114664 Added jgrapht dependency. hanna 2009-10-28 14:29:17 +0000
  • 06ff81efe5 Added NeighborhoodQualityWalker.java and ReadQualityScoreWalker.java which are used to calculate a read quality score based on attributes of the read and the reads in the neighborhood. rpoplin 2009-10-28 13:24:11 +0000
  • 68fa6da788 Initial graph-based reference implementation and alignment assessor. Not suitable for public use depristo 2009-10-27 21:54:47 +0000
  • 31d143a841 now only needs READS depristo 2009-10-27 21:54:14 +0000
  • ef2ea79994 code cleanup and containsStartPosition function depristo 2009-10-27 21:53:40 +0000
  • 186a8dd698 Trivial protection for null value depristo 2009-10-27 21:52:52 +0000
  • be333da9c0 charSeq2byteSeq -- convert a char[] to a byte[] for convenience depristo 2009-10-27 21:52:23 +0000
  • 4192b093b8 More robust error handling with parallelization + usePreviousBase. Added forceReadBasesToMatchRef to use in conjunction with nPreviousReadBases as a less stringent approximation of usePreviousBases (requiring previous pileups only had mismatches, and that read mapping quality be high was throwing everything away) chartl 2009-10-27 17:20:44 +0000
  • 31d5df2859 Previous base now checks that the read matches the reference in the previous base window. chartl 2009-10-27 15:58:20 +0000
  • 726378be8b Almost ready to stop doing eagar decoding; waiting on Eric depristo 2009-10-27 15:28:05 +0000
  • e96b1791ab Need to check for biallelic snp or exception gets thrown. Also, update to new tracker calls. ebanks 2009-10-27 02:43:43 +0000
  • 3fb3773098 a fix for traverse dupplicates bug: GSA-202. Also removed some debugging output from FastaAltRef walker aaron 2009-10-26 20:18:55 +0000
  • a1e8a532ad Support for initialize() and onTraversalDone() output from parallelized walkers. hanna 2009-10-26 20:18:31 +0000
  • 62c1001790 BTTJ is now correct. What a terrible waste of time, turns out I'd just reversed the header. Because of this the MD5 had to be updated in the tests. chartl 2009-10-26 19:24:18 +0000
  • 24c7f694e6 Handles allele frequencies for any specified population, changed user input for mismatch filter options sjia 2009-10-25 22:51:56 +0000
  • db9419df49 @ Hack to allow output from onTraversalDone() chartl 2009-10-25 15:19:04 +0000
  • 75ad6bbef7 Check that map isn't being called passing in null arguments. (This seems wrong; see JIRA entry GSA-211) ebanks 2009-10-25 02:30:36 +0000
  • b4f55df600 Bugfix for Jason F depristo 2009-10-24 22:09:27 +0000
  • 8f381c42fb A location in our repository for archiving files and data that are not currently required, but will be required in the future. hanna 2009-10-23 17:06:33 +0000
  • 65b98470f3 Temporary fix: have RodLocusView manage and close its RODs. Really the relationship between these two classes needs to be rethought; see JIRA GSA-207. hanna 2009-10-23 16:00:12 +0000
  • ad1fc511b1 intermediate commit for some changes in the Variation system, so Eric can go ahead with his changes. Everything is pretty set, but the Variation interface could use a convenience method that joins all the alternate alleles. aaron 2009-10-23 06:31:15 +0000
  • 6c338eccb8 Joint Estimation model now emits calls in all formats. The whole GenotypeCall framework needs to be changed, but this will work for the time being. ebanks 2009-10-23 03:07:28 +0000
  • a6dc8cd44e BTTC is now Tree Reducible allowing for parallelization. chartl 2009-10-22 23:19:29 +0000
  • 2e552eb5a1 Validates intervals against sequence dictionary header bounds. hanna 2009-10-22 19:31:15 +0000
  • 54c61c663c -Cleanup of the Joint Estimation code -Don't print verbose/debugging output to logger, but instead specify a file in the argument collection (and then we only need to print conditionally) ebanks 2009-10-22 15:25:29 +0000
  • 2cab4c68d4 Added method: isCodingExon(). Returns true if position is simultaneously within an exon AND within coding interval of any single transcript from the list. The old method of detecting coding positions as isExon() && isCoding() is buggy, as the position could be in the UTR part of one transcript (isExon() is true), and within coding region bounds (but not in the exon) of another transcript (isCoding() is true). As a result UTR positions would be erroneously annotated as coding. asivache 2009-10-22 14:55:07 +0000
  • af761fb9bd Base transition table now forces epsilon/3 (three-state) model for the unified genotyper. Verified to be identical with changing the default model to being epsilon/3. This of course changes the observed counts, so the integration test has been updated. chartl 2009-10-21 21:18:26 +0000
  • 55fa1cfa06 -Renamed new calculation model and worked out some significant xhanges with Mark -Allow walkers calling the UG to pass in their own argument collections ebanks 2009-10-21 20:49:36 +0000
  • 8e3f72ced9 BTTJ - Code refactoring (major) - passes integration test chartl 2009-10-21 19:04:51 +0000
  • 15a1849758 notes for chartl depristo 2009-10-21 18:31:31 +0000
  • 77863d4940 @PowerBelowFrequency + Changes to doc chartl 2009-10-21 15:17:11 +0000
  • 967128035e Make command like args default to false. chartl 2009-10-21 13:59:35 +0000
  • 9b9744109c Mark's new unified calculation model is now officially implemented. Because it doesn't actually use EM, it's no longer a subclass of the EM model. ebanks 2009-10-21 02:39:23 +0000
  • caa3187af8 Enabling correct high-performance ROD walker and moved VariantEval over to it. Performance improvements in variantEval in general. See wiki for full description depristo 2009-10-20 23:31:13 +0000
  • 4a8a6468be Use read group as a condition for confusion tables. With an integration test. chartl 2009-10-20 19:39:32 +0000
  • b83df5616a Change for lower-case references (always compare upper case bases) chartl 2009-10-20 17:36:31 +0000
  • 3b1fabeff0 Major code refactoring: chartl 2009-10-20 14:58:04 +0000
  • 4be6bb8e92 added a check to ensure the eval track variation is bi-allelic. Also changed some string constants over to enums. For some reason my check-ins from home wouldn't work last night, so this is the actual changes for 1884. aaron 2009-10-20 14:15:33 +0000
  • 449a6ba75a Deleting lots of code as part of my cleanup. More classes tagged for removal. Many more walkers have their days numbered. depristo 2009-10-20 12:23:36 +0000
  • d749a5eb5f added a check to ensure the eval track variation is bi-allelic. Also changed some string constants over to enums aaron 2009-10-20 04:56:51 +0000
  • b8ab77c91c Don't filter out reads without proper read groups. Instead, allow the user (or another walker calling UG) to specify an assumed sample to use (but then we assume single-sample mode). ebanks 2009-10-20 01:30:53 +0000
  • a8a2c1a2a1 Replaced SSG with UG in packaging utils. Minor performance and formatting improvements for ClipReads depristo 2009-10-20 01:19:58 +0000
  • c29924e7cf Reverting previous change. Aaron, it's all yours... ebanks 2009-10-20 00:55:24 +0000
  • d21b582b18 memory leak, where the Resource Pool was releasing based on the value and not the key, resulting in the resourceAssignments map growing with each additional shard aaron 2009-10-20 00:39:42 +0000
  • 761a730758 assertBiAllelic -> assertMultiAllelic. Chris, if this breaks an integration test, you get it. ebanks 2009-10-20 00:09:46 +0000
  • 2a26bb42dd Softclipping support in clip reads walker. Minor improvement to WalkerTest -- now can specify file extensions for tmp files. Matt -- I couldn't easily create non-presorted SAM file. The softclipper has an impact on this. depristo 2009-10-19 21:54:53 +0000
  • 055a99fb05 Change in ordering for a disjunctions. Walker will no longer try to calculate number of simple mismatches in the pileup if the pileup includes 'N's. chartl 2009-10-19 18:24:14 +0000
  • 10bde9e77b Integration test for BTT calculator chartl 2009-10-19 18:21:55 +0000
  • cfa86d52c2 ensure that in the indel case we don't allow identification as both an insertion and deletion at the same location in the VCF ROD aaron 2009-10-19 18:21:00 +0000
  • 3d50c72d74 Forgot a dumb little System.out.println. You will be flooded with "This read will not be used." statements until, overwhelmed, you give in to my demands. chartl 2009-10-19 16:13:48 +0000
  • 225ef52973 Now produces same output as the Scala walker for unconditioned tables (no 2bb, no previous base, etc.) chartl 2009-10-19 16:10:44 +0000
  • f0021a3533 Changes to the BTTC scala walker: chartl 2009-10-18 18:09:25 +0000
  • bb180a23ef Updated MD5 ebanks 2009-10-18 05:30:38 +0000
  • 51f9ec0a5c subtract largest posterior value from all values; this hopefully solves any precision issues ebanks 2009-10-18 05:20:15 +0000
  • b9e8867287 -push allele frequency and genotype likelihood variable definitions down into the subclasses so that they can use different data structures -use slightly more stringent stability metric -better integration test ebanks 2009-10-18 04:22:17 +0000
  • d6385e0d88 simpleComplement function() in BaseUtils. Generic framework for clipping reads along with tests. Support for Q score based clipping, sequence-specific clipping (not1), and clipping of ranges of bases (cycles 1-5, 10-15 for example). Can write out clipped bases as Ns, quality scores as 0s, or in the future will support softclipping the bases themselves. depristo 2009-10-16 22:29:35 +0000
  • ad777a9c14 @BasicPileup - made the counts public so they can be used chartl 2009-10-16 21:56:56 +0000
  • bdb34fcf38 Updated integration tests for VariantEval. Hooray for IT! andrewk 2009-10-16 20:00:29 +0000
  • 41a95cb3f0 fixing unified genotyper test for change: VCF output now emits no calls as ./. aaron 2009-10-16 19:38:58 +0000
  • 85a4fbc256 Bumping version of Picard for firehose compatibility. Integration tests were validated against svn rev 1861, before the wonder twins committed their changes. hanna 2009-10-16 19:38:56 +0000
  • 8aacc43203 VCF output now emits no calls as ./. aaron 2009-10-16 18:51:31 +0000
  • d1a4cd2f73 Added ValidationData analysis type to VariantEvalWalker; this eval takes a GFF file with validated truth data positions (bound to "validation")and calculates the accuracy of the genotype calls bound to "eval". andrewk 2009-10-16 15:39:08 +0000
  • 07b134a124 Added some integration tests for multiple samples ebanks 2009-10-16 15:22:10 +0000
  • 418e007ca6 A cleaner interface: now everyone can use UG's initialize method ebanks 2009-10-16 14:09:16 +0000
  • 96972c3a5c a fix for a bug Eric found: if your first call contains fewer samples than calls at other loci, your VCFHeader got setup incorrectly. aaron 2009-10-16 04:57:50 +0000
  • a69ea9b57c Cleaning up the VCF code, adding lots of tests for a variety of edge cases. Two issues are still outstanding: updating the no call string with the standard 1000g decided on today, and fixing Eric's issue where not all the VCF sample names are present initially. aaron 2009-10-16 04:11:34 +0000
  • b82c3b6040 Better error output (and fixed spelling mistakes) ebanks 2009-10-16 01:01:45 +0000
  • 993c567bd8 I had to remove some of my more agressive optimizations, as they were causing us to get slightly different results as MSG. Results in only small cost to running time. ebanks 2009-10-16 00:59:32 +0000
  • 7d7ff09f54 throw an exception if read has no associated read group asivache 2009-10-15 18:11:32 +0000
  • b9544d3f89 Output formatting change (very slight) chartl 2009-10-15 16:47:29 +0000
  • 839c5d66bc Read uints directly into longs. hanna 2009-10-15 16:15:11 +0000
  • ce38fa7c81 Breaking the signed int glass ceiling; stage 1: convert critical ints to longs. Code cleanup and documentation. hanna 2009-10-15 15:28:56 +0000
  • 79993be46c changed blank gene name to UNKNOWN kcibul 2009-10-15 13:47:00 +0000
  • 0c2016c19a Improved error messages -- now easier to read, points to the GATK Error Messages wiki, and avoids double printing of stack traces depristo 2009-10-15 12:07:44 +0000
  • a9094c835c clean-up and fixes to the VCF input aaron 2009-10-15 04:53:59 +0000
  • a32470cea1 Deal with the fact that walkers can call UG's init/map functions directly. We need to filter contexts in that case since the calling walkers don't get UG's traversal-level filters. ebanks 2009-10-15 02:31:45 +0000