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

Commit Graph

  • b05119987c Clarified some of the comments in the individual covariates now that things have been moved around to speed up the code. In general most error checking and adjustments to the data are done per read instead of per base. This means that functionality was moved out of the covariate modules and into CovariateCounterWalker and TableRecalibrationWalker. rpoplin 2009-11-14 18:44:05 +0000
  • 672472789e Added some documentation to the helper classes. Fixed an error case in TableRecalibrationWalker. rpoplin 2009-11-14 18:13:43 +0000
  • 15c14add4d Repackage the aligner for better partitioning. The C aligner, for example, is now partitioned from the Java aligner, and both are partitioned from the more general- purpose BWT reader. hanna 2009-11-13 22:55:27 +0000
  • b7097a8db9 Build aligner as separate jar file. hanna 2009-11-13 22:29:29 +0000
  • f5818c0cda Fix max edit distance to ensure interface similarity between our aligner and 'bwa aln'. hanna 2009-11-13 22:18:56 +0000
  • 52494d8176 cleanup of SNP selector -- ready for some additional testing depristo 2009-11-13 21:46:31 +0000
  • 8eff1cc436 Extract and include only the Tim Fennell-approved parts of picard private. Hopefully this is a temporary solution and these classes will be migrated into picard-public. hanna 2009-11-13 19:42:33 +0000
  • d1b525b428 Default window size for NQS covariate is 3 rpoplin 2009-11-13 19:24:27 +0000
  • 394c839974 Implemented NQS covariate. Extended Cycle covariate to handle 454 and SOLID reads. Added a Primer Round covariate for SOLID reads. rpoplin 2009-11-13 19:22:21 +0000
  • bf451873ff 1. Bug fix: check that AF=0 doesn't contain more probability than 1-fraction 2. Fix for Kiran: allow UG to call SNPs at deletion sites; we'll add an annotation to the VariantAnotator for deletions at the locus (next week). 3. Added integration tests for joint estimation model ebanks 2009-11-13 18:02:18 +0000
  • 1be36ca959 Bug fix: when cleanedReadIterator is initialized, it gets immediately set to the contig of the first cleaned read; when the first uncleaned read coming in is on the lower contig, this would trigger 'readNextContig' with that lower contig as an argument. As the result, the whole cleaned reads file would be read through the end and no cleaned reads would be ever seen by the code afterwards. Now we do not call readNextContig if the (uncleaned) read's contig is lower than the current contig already loaded into cleanedReadIterator. the 'readNextContig' method now also throws an exception if requested contig is less than the currently loaded one asivache 2009-11-13 15:41:26 +0000
  • b1376e4216 structure refactored throughout for performance improvements rpoplin 2009-11-13 15:41:09 +0000
  • cff31f2d06 comments for eric depristo 2009-11-13 14:19:31 +0000
  • 234bb71747 changed the toVariation() method to take a reference base, instead of using the reference base loaded from the underlying data source (if it was reference aware). Also changed some isVariant() methods which weren't using the passed in ref base. aaron 2009-11-13 06:54:38 +0000
  • 902cf84448 Bug fix: if the most likely allele frequency is 0, don't make a variant call (even if the Qscore for AF=1/n > threshold) ebanks 2009-11-13 04:10:32 +0000
  • 555fb975de 1. Print out allele frequency range (from joint estimation model only). 2. Don't print verbose output from SLOD calculation (it's just a repeat of previous output). ebanks 2009-11-13 03:59:13 +0000
  • 72825c4848 A walker that generates a table of secondary base counts in a bam file. mmelgar 2009-11-13 02:11:23 +0000
  • 7c386fa428 Another case of reordering of read groups blowing up checksums. hanna 2009-11-13 00:07:35 +0000
  • 8145ed4672 Take 2, updating picard with bug fix for bam files containing no reads. Just stomped on the existing md5s because that's what Eric told me to do. hanna 2009-11-12 22:52:08 +0000
  • 61b5fb82ce 2 major changes: ebanks 2009-11-12 22:51:49 +0000
  • 3742a05760 Now can read E2 or SQ tag. mmelgar 2009-11-12 15:18:21 +0000
  • c3c001e02e cleanup of the traversal output code aaron 2009-11-12 06:18:10 +0000
  • 0922400ca9 Don't try to calculate ratios when DoC is zero (which happens when calls are made by an LD-aware genotyper) ebanks 2009-11-12 02:51:44 +0000
  • 697d7e02c8 Remove the lazy initialize functionality. When no calls are made by the genotyper, we still want a vcf file to be output with valid header. ebanks 2009-11-12 02:14:50 +0000
  • 2ea85fb62b Fix some problematic command-line argument naming and descriptions. hanna 2009-11-12 02:12:26 +0000
  • 1a4d071d37 Better snpSelector, plus VCFmerge tool depristo 2009-11-11 22:02:57 +0000
  • 0c2a957ae0 Better configuration support. Now supports everything that people have expressed interest in except edit distance. hanna 2009-11-11 20:54:49 +0000
  • 6c9f86bb4d Removed unnecessary output and added debugging print() routine depristo 2009-11-11 18:37:36 +0000
  • 637f02f392 BWA/C changes have arrived in svn repository. Remove temporary storage from Sting repository. hanna 2009-11-11 16:48:21 +0000
  • 578dcc54a4 Don't create a record if ref=N ebanks 2009-11-11 04:32:17 +0000
  • 8406325247 New Picard is breaking one of the integration tests. Revert until we find out whether the cause is legit. hanna 2009-11-11 03:59:32 +0000
  • 499e7d1d75 Push forward some more delicate merging routines. hanna 2009-11-11 03:07:34 +0000
  • bae4d3f7ea Updated Picard with fix for Doug Voet. Thanks Alec. hanna 2009-11-11 02:01:08 +0000
  • 2e4782f202 Command-line arguments for SamReadFilters. hanna 2009-11-10 23:36:17 +0000
  • a13cbe1df0 The refactored recalibrator now passes the integration tests as well as my own validation tests. I'm ready to have other people start jamming on the files. I'll make an updated wiki page soon. The refactored recalibrator is currently a bit slower than the old one but there were a lot of great, easy ideas today for how to improve it. rpoplin 2009-11-10 22:20:06 +0000
  • 2cf9670d1e Allow users to directly specify filters from the command-line, applicable to any walker. hanna 2009-11-10 18:40:16 +0000
  • 6a37090529 Output changes for VCF and UG: 1. Don't cap q-scores at 99 2. Scale SLOD to allow more resolution in the output 3. UG outputs weighted allele balance (AB) and on-off genotype (OO) info fields for het genotype calls (works for joint estimation model and SSG) ebanks 2009-11-10 16:31:31 +0000
  • 1e7ddd2d9f Added a validateOldRecalibrator option to CovariateCounterWalker which reorders the output to match the old recalibrator exactly. This facilitates direct comparison of output. Changed the -cov argument slightly to require the user to specify both ReadGroupCovariate and QualityScoreCovariate to make it more clear to the user which covariates are being used. Some speed up improvements throughout. rpoplin 2009-11-10 15:55:56 +0000
  • 7e30fe230a oops, missing file depristo 2009-11-10 13:25:18 +0000
  • d316cbad4c VariantFilteration now accepts a VCF rod in addition to an input geli. It will then annotate this VCF file with filtering information in the INFO field too. --OnlyAnnotate will not write in filtering output depristo 2009-11-10 13:24:58 +0000
  • f9819d5f13 a little clean-up aaron 2009-11-10 06:18:34 +0000
  • 2ed423ed56 print the current location in read walkers (in addition to the number of reads processed), along with some refactoring to support the change. aaron 2009-11-10 05:57:01 +0000
  • c9c3cf477a Based on feedback from Kiran, we know uniquify sample names as sample.rodName (instead of sample.1, sample.2, ...) ebanks 2009-11-10 02:41:37 +0000
  • 3990c6d950 snpSelector v3 -- bootstrapping support and VCF output depristo 2009-11-09 22:48:51 +0000
  • 2fa2ae43ec Enough people have found this useful, so... Moving Callset Concordance tool to core and adding integration test. ebanks 2009-11-09 20:59:18 +0000
  • 3793519bd4 -Added convenience method to VCF record to tell if it's a no call and have rodVCF use it before querying for info fields -Don't restrict info fields to 2-letter keys [about to move these to core] ebanks 2009-11-09 20:52:51 +0000
  • 740a5484c4 Added some documentation to the code, mostly especially to CovariateCounterWalker but various comments added throughout. Also changed the HashMap data structure to accept an estimated initial capacity. This had a very modest improvement to the speed. rpoplin 2009-11-09 20:13:56 +0000
  • 74751a8ed3 -Some minor fixes to get accurate vcf record merging done -Improvement to snp genotype concordance test ebanks 2009-11-09 06:40:55 +0000
  • ab705565cf Completely refactored the Callset Concordance code. Now, it takes in VCF rods and emits a single VCF file which has merged calls from all inputs and is annotated (in the INFO fields) with the appropriate concordance test(s). Still needs a bit of polish... ebanks 2009-11-09 05:03:13 +0000
  • bc6f24e88f Added VCFUtils which contains some useful VCF-related functions (e.g. ability to merge VCF records). Also, various minor improvements. ebanks 2009-11-09 04:53:32 +0000
  • cff645e98b convenience method to deal with genotypes that are unsorted (e.g. CA vs. AC) ebanks 2009-11-09 04:45:49 +0000
  • 7fde6c0bf4 One more output tweak. kiran 2009-11-09 04:42:55 +0000
  • 00a7113d7a Tweaks to formatting of output table. kiran 2009-11-09 04:33:36 +0000
  • 7ce0df76f8 Added accessors to the rod data sources so that walkers can access the name/file/type triplets for input rods. This is necessary if e.g. you want to create a vcf writer based on all of the samples being input. ebanks 2009-11-09 04:25:39 +0000
  • d07f3bb6f6 Added methods to get strand bias and to test if record has allele freq or bias fields set. ebanks 2009-11-09 04:20:35 +0000
  • 3313b0ddb4 Fixed a minor bug where the lodThreshold wasn't being printed in the header. kiran 2009-11-08 16:51:36 +0000
  • 95d381efe2 Optionally computes the error rate using the best base and a random base. kiran 2009-11-08 16:47:34 +0000
  • 567f5758d2 Optionally lists read depths by read group. kiran 2009-11-08 16:39:19 +0000
  • a679bdde18 FindContaminatingReadGroupsWalker lists read groups in a single-sample BAM file that appear to be contaminants by searching for evidence of systematic underperformance at likely homozygous-variant sites. kiran 2009-11-08 16:36:39 +0000
  • 2225d8176e A convenience class for maintaining a dynamically growing table of values with access to the elements by named row and column identifiers. kiran 2009-11-08 16:34:35 +0000
  • 21c5f543fa Fix sharding bug -- loci to which >100,000 (= 1 shard) reads are assigned an alignment start will confuse the sharding system and cause it to return duplicate reads. hanna 2009-11-08 14:27:26 +0000
  • f777c806d6 snpSelector v2 -- code refactoring and support for comparison with known truth. Looks great. depristo 2009-11-07 19:32:12 +0000
  • 84ba604611 Sequential quality score calculation is now in place in the refactored recalibrator and matches the quality scores calculated by the old recalibrator exactly; at least on the small sets of data used so far. Validation, documentation, and optimization work is on going. rpoplin 2009-11-07 15:55:16 +0000
  • bf1bc94060 Fixes for PooledConcordance bugs and lack of safety checking depristo 2009-11-07 01:54:10 +0000
  • 7cb51dbc31 snpSelector v1 -- and supporting changes to VCF reader depristo 2009-11-06 23:00:46 +0000
  • 66d4a995e6 Initial check in of refactored Recalibrator. The new walkers are called CountCovariatesRefactored and TableRecalibrationRefactored. More work is needed to finish up the sequential calculation and to document the code sufficiently. These files are not ready to be used by other people quite yet. rpoplin 2009-11-06 22:33:55 +0000
  • 6fdfc97db6 Added optional field DP to VCF output for Mark. ebanks 2009-11-06 20:03:22 +0000
  • f73dd09399 More preparation for bwa patch: create a crude, minimal build system to build against a static library that the bwa compile (now) generates. hanna 2009-11-06 18:41:01 +0000
  • 0a55fa5bb1 Completely refactored the Genotype Concordance module(s). Now PooledConcordance and GenotypeConcordance inherit from the same super class (and can therefore share data structures and functionality). Also, they now use ConcordanceTruthTable to keep track of necessary info. GenotypeConcordance passes integration tests. PooledConcordance needs to be finished by Chris. ebanks 2009-11-06 16:27:16 +0000
  • d549347f25 Refactored GenotypeLikelihoods to use an underlying 4-base model. It needs to be modified a bit and then hooked up to a pooled model, but that is now possible. At this point, there is no difference to the Unified Genotyper. ebanks 2009-11-05 21:59:25 +0000
  • 4d3871c655 don't flush anymore. jmaguire 2009-11-05 19:11:51 +0000
  • aacd72854f a fix for a bug Andrey discovered: in read-based interval traversals we're dupplicating reads in rare cases. The problem was that to accomidate a bug in SAM JDK indexing, we were forced to add one to the stop of our QueryOverlapping() calls to ensure we always got all of the overlapping reads. aaron 2009-11-05 05:26:33 +0000
  • 15e80aec3b Temporarily adding changed bwa files to Sting repository until I can pull together the official bwa patch. hanna 2009-11-04 21:46:05 +0000
  • eca0942644 Oops. Let's make sure only to write calls that the pool supports to the auxiliary vcf files. chartl 2009-11-04 17:14:55 +0000
  • 43c3ee61d5 Fix minor mapping quality bug. hanna 2009-11-04 14:33:23 +0000
  • 21b1ece45d removed some extra debugging commands from the build.xml aaron 2009-11-04 06:28:12 +0000
  • fc17e75759 Put this puppy through its paces. Eliminated the sorting and header-handling stuff; that isn't the purvey of this script and should be handled downstream or by a script wrapper. chartl 2009-11-04 06:05:45 +0000
  • 3d9195f8b6 Added - converter from expanded summary to VCF (beautiful thing, really) chartl 2009-11-03 22:20:47 +0000
  • a545859c62 Joint Estimation model now emits a reasonable slod ebanks 2009-11-03 21:12:42 +0000
  • 11d950abe0 No longer allow the lod_threshold argument - use confidence instead. Have UG output qscores in all cases. ebanks 2009-11-03 16:18:51 +0000
  • 2fb45dbd73 Make window size a command line argument asivache 2009-11-03 16:13:35 +0000
  • 55f61b1f88 Bug fix in adjustment of the shift position. asivache 2009-11-03 16:08:11 +0000
  • d60c632099 Minor output improvement depristo 2009-11-03 13:20:55 +0000
  • 44ea55d338 Useful library for parsing VCF files, plus a general VCF->table converter depristo 2009-11-03 13:14:04 +0000
  • 5d5dc989e7 improvements to VCF and variant eval support of VCF -- now listens to the filter field depristo 2009-11-03 12:09:30 +0000
  • c313abf0be Restore a jar packaging change that got clobbered. hanna 2009-11-03 00:15:45 +0000
  • c63af32fc7 The BWA/C bindings were triggering the local aligner to repeatedly reload the ref genome. Make sure the reference genome is cached. hanna 2009-11-03 00:01:55 +0000
  • 3a33401822 2nd stage of the genotyper output refactoring is complete. Now, all output is generalized and all of the intelligence lies where it is supposed to. Next stage is syncing up old and new models and making sure we're outputting exactly what we should. ebanks 2009-11-02 22:43:08 +0000
  • ba67c7f02b added a warning for those using bed files; we properly convert bed to the internal representation but the user needs to be aware that any output will be one-based closed intervals aaron 2009-11-02 21:09:18 +0000
  • b71b66bd88 the underlying parameter is a float so we need to use Float.valueOf() instead; Noticed by external user Hou Huabin aaron 2009-11-02 20:22:25 +0000
  • 5a510e6d98 New PackageUtils interferes with the packaging utility. Revert until Aaron and I can get together to make this work. hanna 2009-11-02 19:14:14 +0000
  • de6ae51f7e Scala walkers can now be build and run like any other walker in the GATK. Added the getUrlsForClasspath to PackageUtils, the Reflections package isn't getting the manifest files from jars in the classpath, and so we weren't seeing any walkers outside of the GenomeAnalysisTK.jar. aaron 2009-11-02 06:02:41 +0000
  • 1896f334d9 Fixed collection of bugs in reads aligning to multiple locations. hanna 2009-11-02 04:02:09 +0000
  • af6d0003f8 -Generalized the GenotypeConcordance module to deal with any number of individuals (although it will default to its old behavior if the -samples argument is left out). -Make rods return the appropriate type of Genotype calls from getGenotype(). ebanks 2009-11-01 05:35:47 +0000
  • b95165e39c Make alignment (temporarily) part of main GenomeAnalysisTK.jar. Add some extra logging errors on failure. hanna 2009-11-01 00:33:18 +0000
  • 4b0796ba58 After fixing a few glitches and bugs, this version finally works as intended asivache 2009-10-31 04:59:58 +0000
  • 7d0ac7c6f2 Fix for long-term VariantEval bug plus new intergration test to catch it depristo 2009-10-31 00:00:33 +0000
  • 99337df929 Now looks up and propagates Syzygy's LOD scores into the appropriate field (so variantfiltration can adjust lod scores accurately) chartl 2009-10-30 21:13:03 +0000
  • ea8d5c7077 Some internal refactoring. Now "safely" ignores duplicate records (NOT duplicate reads but rather malformed bam files!) resulting from the bug/feature in CleanedReadInjector. asivache 2009-10-30 17:50:51 +0000
  • 7654051aee Faster grepping chartl 2009-10-30 16:59:17 +0000