40797f9d45
Ensuring a minimum number of variants when clustering with bad variants. Better error message when Matrix library fails to calculate inverse.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5793 348d0f76-0448-11de-a6fe-93d51630548a
2011-05-12 01:48:37 +00:00
6c7a0adc76
Updating VariantGaussianMixtureModelUnitTest to use truth sensitivity cutting
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5750 348d0f76-0448-11de-a6fe-93d51630548a
2011-05-04 13:56:01 +00:00
23cd3a7a5d
Moving VQSR v2 to core.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5740 348d0f76-0448-11de-a6fe-93d51630548a
2011-05-03 20:20:06 +00:00
ff7edc4493
Minor bug fix in empiricalMu prior calculation in VQSR.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5412 348d0f76-0448-11de-a6fe-93d51630548a
2011-03-10 00:42:38 +00:00
509daac9f7
Minor bug fix in k-means implementation. Updating VQSR integration tests in preparation for VQSRv2 by removing some unused features such as VariantDatum.weight and ti/tv cutting.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5410 348d0f76-0448-11de-a6fe-93d51630548a
2011-03-09 00:26:28 +00:00
01323447c6
Removed LibBat.SUB2_BSUB_BLOCK since the use of it exits the JVM.
...
Fixed integration tests to wait on their own for the job to run instead of using SUB2_BSUB_BLOCK.
Updated VariantRecalibrationIntegrationTests MD5s which were knocked out of sync whele SUB2_BSUB_BLOCK was exiting in the middle of integration tests.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4840 348d0f76-0448-11de-a6fe-93d51630548a
2010-12-14 19:57:20 +00:00
abd6ce1c77
A TiTv-free approach for cutting variants! Apparently much better than previous approach, and will work for indels and SV will truly minor modifications to the code. Will discuss with methods group on Monday.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4822 348d0f76-0448-11de-a6fe-93d51630548a
2010-12-11 23:08:13 +00:00
f1f01610f8
Remove the extra trailing tab at the end of the VCF ## header line. Unfortunately, this meant updating every freaking integration test.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4806 348d0f76-0448-11de-a6fe-93d51630548a
2010-12-08 17:22:29 +00:00
0adf505b53
We no longer look at by-hapmap validation status in the VQSR because using the HapMap VCF file is higher quality. As a side effect we now support the dbsnp 132 vcf file. ApplyVariantCuts now requires that the input VCF rod bindings begin with input, matching the other VQSR walkers. Wiki updated with information about how to obtain the hapmap and 1kg truth sets.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4772 348d0f76-0448-11de-a6fe-93d51630548a
2010-12-02 15:38:45 +00:00
24ec35deaf
- Reintroduce test dependency so that the tests passing / failing is not
...
dependent on the contents of the integrationtest directory. Will figure
out how to better manage the integrationtest directory at some point in
the future.
- Up the max heap size for tests.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4691 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-16 19:55:20 +00:00
ef2f6d90d2
VQSR now operates on LOD scores in the INFO field directly, and doesn't adjust the QUAL field. New format for tranches file uses LOD score. Old file format no longer supported. log10sumlog10() function, a very useful utility in MathUtils. No more ExtendedPileupElement! Robust math calculations in GMM so that no infinities are generated! HaplotypeScore refactored to enable use of filtered context. Not yet enabled... InferredContext getDouble and getInteger arguments now parse values from Strings if necessary
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4684 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-15 22:19:22 +00:00
2fd816ac5f
Updated ordering of integration tests. GVC > VR > AVC
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4669 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-14 06:33:28 +00:00
44d0cb6cde
New version of cutting routines for VQSR. Old code removed. Working unit tests. Best practice with testng integration test (everyone look at it). Walker test now allows you to not specify no. input files, if it can infer input counts from MD5s
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4664 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-13 16:19:56 +00:00
62a106ca5a
Disabled VariantGaussianMixtureModelUnitTest
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4663 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-13 03:53:33 +00:00
42acc968b1
Unit tests
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4660 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-12 20:09:39 +00:00
988da428ae
Bug fix for old style tranches file. ApplyVariantCuts moved over, and passes integration tests
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4657 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-12 14:38:26 +00:00
861ee3e37a
Changing testing framework from junit -> testng, for its enhanced configurability.
...
Initial test to see how Bamboo will respond. More detailed email to follow.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4609 348d0f76-0448-11de-a6fe-93d51630548a
2010-11-01 21:31:44 +00:00
0a2e76e9dc
2nd step towards on the fly indexing. Also fixed parsing bug for headers with < symbols
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4454 348d0f76-0448-11de-a6fe-93d51630548a
2010-10-07 21:38:46 +00:00
0de658534d
Removed the qScale arguments in VariantRecalibrator. It is smarter about how it tries to find a cut so the arbitrary scale factor hopefully is no longer necessary. Now the recalibrated variant quality score more accurately reflects our believed lod of the call.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4451 348d0f76-0448-11de-a6fe-93d51630548a
2010-10-07 18:04:57 +00:00
69485d6a7a
Added command line argument for the max value of the allele count prior in VariantRecalibrator (--max_ac_prior). Default value increased to 0.99 from 0.95.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4436 348d0f76-0448-11de-a6fe-93d51630548a
2010-10-06 14:00:53 +00:00
a6c7de95c8
By using the AC info field instead of parsing the genotypes we cut 78% off the runtime of VariantRecalibrator. There is a new argument to force the parsing of genotypes if necessary. Various other optimizations throughout.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4383 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-29 18:56:50 +00:00
547763b230
Better error message for Petr's null pointer exception. Also added an exception integration test because I'm certain this used to work.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4319 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-21 13:44:40 +00:00
0a06fbdb94
Adding header lines to output of VR walkers to settle validator warnings. Command lines are added to the VCF header. GATK version numbers will be added to the header lines by Matt.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4288 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-15 16:45:03 +00:00
4e83ba411f
We now do lazy loading for the genotype data in VCF. Practically, almost all walkers end of loading the genotype data because we need to be smarter about transfering the unparsed genotype string when modifying VariantContexts; however, this does solve the problem for VR's piece to generate clusters (shaved off 75% of runtime for Ryan's large case). That further optimization will happen later.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4279 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-15 00:18:17 +00:00
7b113a4886
Truncate the floating point numbers coming out of the variant recalibration walkers. Integration tests now work with both 1.6.0_16-b01 and 1.6.0_21-b06
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4253 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-10 18:37:49 +00:00
61e848c4f0
It's clear from Sendu's calling and my own calling that -qScale 100.0 is a much better default value for low pass data.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4248 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-10 01:47:21 +00:00
aeb897db7f
VR walkers look at by-hapmap validation status by default. Eric will be updating the syntax to allow for more flexibility here.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4242 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-09 15:40:56 +00:00
d625186796
I think the VR integration tests are fine.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4240 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-09 15:00:41 +00:00
6a30617a60
Initial implementation of UserError exceptions and error message overhaul. UserErrors and their subclasses UserError.MalFormedBam for example should be used when the GATK detects errors on part of the user. The output for errors is now much clearer and hopefully will reduce GS posts. Please start using UserError and its subclasses in your code. I've replace some, but not all, of the StingExceptions in the GATK with UserError where appropriate.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4239 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-09 11:32:20 +00:00
e3962c0d13
VR integration tests are longer but much more useful.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4210 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-03 15:50:19 +00:00
0bb05fb472
Bug fix in VariantRecalibrator. Only add sample names from the input rod bindings, not from all rod bindings.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4194 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-02 21:12:09 +00:00
469bbaa240
Added more integration tests for the variant quality score recalibrator
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4181 348d0f76-0448-11de-a6fe-93d51630548a
2010-09-01 15:31:24 +00:00
9c3f403307
Add the calculated lod value to the info field of each recalibrated VCF record.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4153 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-27 21:33:58 +00:00
5623e01602
GenerateVariantClusters and VariantRecalibrator now uses hapmap and 1kg ROD bindings (in addition to dbsnp) to distinguish between knowns and novels. It no longer looks at by-hapmap validation status so providing hapmap is highly recommended. Example on the wiki. Input variants tracks now must start with input.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4113 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-25 18:33:40 +00:00
bf0b6bd486
Update integration tests to use the new ROD syntax.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4112 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-25 18:13:30 +00:00
85007ffa87
Some clean up for the variant recalibrator. Now uses @Input and @Output so that it can join the Queue party. Users now specify a -o, -clusterFile, -tranchesFile, and -reportDatFile. Example on the wiki. ApplyVariantCuts now has an integration test. Base quality recalibrator now requires a dbsnp rod or vcf file. Now that the base quality recalibrator is using @Output the PrintStream shouldn't be closed in OnTraversalDone.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4101 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-24 20:14:58 +00:00
8f15b2ba72
Memory optimization for the VariantRecalibrator. Only add variants to the list if they pass the novelty and qual filters.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4051 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-17 21:57:28 +00:00
578e7fa36d
Don't output -0 as qual value in VariantRecalibrator.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4044 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-16 16:47:58 +00:00
3eee3183fd
Checking in the tiger team changes. LOD calculation modified. -qScale is back in case people need it.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3990 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-09 20:41:03 +00:00
a8d37da10b
Checking in everyone's changes to the variant recalibrator. We now calculate the variant quality score as a LOD score between the true and false hypothesis. Allele Count prior is changed to be (1 - 0.5^ac). Known prior breaks out HapMap sites
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3952 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-05 14:12:19 +00:00
340bd0e2c1
Removed hard-coded pointers to references
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3934 348d0f76-0448-11de-a6fe-93d51630548a
2010-08-04 17:59:37 +00:00
c6ad26e04f
1) When quals/GQs are really integers (x.00), strip off the floating points.
...
2) Keep track of whether vcf records are unfiltered vs. pass filters in the variant context so we can regenerate the records on output.
3) No more "ID" hard-coded all over the code to set the VariantContext ID. Use a static variable instead.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3840 348d0f76-0448-11de-a6fe-93d51630548a
2010-07-20 18:01:45 +00:00
55b756f1cc
First step in major cleanup/redo of VCF functionality. Specifically, now:
...
a) VCF track name can work again with 3.3 or 4.0 VCF's when specifying -B name,VCF,file. Code will read header and parse automatically the version.
b) Old VCF codec is deprecated. Reader goes now direct from parsing VCF lines into producing VariantContext objects, with no intermediate VCF records. If anyone can't resist the urge to still input files using the old method, a new VCF3Codec is in place with the old code, but it will be eventually deleted.
c) VCF headers and VCF info fields no longer keep track of the version. They are parsed into an internal representation and will be output only in VCF4.0 format.
d) As a consequence, the existing GATK bug where files are produced with VCF4 body but VCF3.3 headers is solved.
e) Several VCF 4.0 writer bugs are now solved.
f) Integration test MD5's are changed, mostly because of corrected VCF4.0 headers and because validation data mostly uses now VCF4.0.
g) Several VCF files in the ValidationData/ directory have been converted to VCF 4.0 format. I kept the old versions, and the new versions have a .vcf4 extension.
Pending issues:
a) We are still not dealing with indels consistently or correctly when representing them. This will be a second part of the changes.
b) The VCF writer doesn't use VCFRecord but it does still use a lot of leftovers like VCFGenotypeEncoding, VCFGenotypeRecord, etc. This needs to be simplified and cleaned.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3813 348d0f76-0448-11de-a6fe-93d51630548a
2010-07-16 22:49:16 +00:00
6b5c88d4d6
The GATK no longer writes vcf3.3; welcome to the world of vcf4.0. Needed to fix a few output bugs to get this to work, but it's looking great. Much more still to come. Guillermo: hopefully this doesn't break your local build too badly.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3786 348d0f76-0448-11de-a6fe-93d51630548a
2010-07-14 04:56:58 +00:00
255b036fb5
Variant Recalibrator MLE EM algorithm is moved over to variational Bayes EM in order to eliminate problems with singularities when clustering in higher than two dimensions. Because of this there is no longer a number of Gaussians parameter. Wiki will be updated shortly with new recommended command.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3704 348d0f76-0448-11de-a6fe-93d51630548a
2010-07-01 18:51:07 +00:00
724affc3cc
Major bug fixes for the Variant Recalibrator. Covariance matrix values are now allowed to be negative. When probabilities are multiplied together the calculation is done in log space, normalized, then converted back to real valued probabilities. Clustering weights have been changed to only use HapMap and by-1000genomes sites. The -nI argument was removed and now clustering simply runs until convergence. Test cases seem to work best when using just two annotations (QD and SB). More changes are in the works and are being evaluated. Misc fixes to walkers that use RScript due to CentOS changes.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3590 348d0f76-0448-11de-a6fe-93d51630548a
2010-06-18 17:37:11 +00:00
b978d5946b
adding changes for VCF 4, mostly in the way we handle VCF headers. The header fields are now aware of the differences between different VCF formats. There was also a bunch of clean-up of out-of-spec VCF used in the tests (mismatched VCF file format fields, etc), and updates to the associated integration tests. Also some logging statements for BTI.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3584 348d0f76-0448-11de-a6fe-93d51630548a
2010-06-18 08:23:23 +00:00
4e268ef6ac
Removing the Variant Recalibration Performance test because it isn't ready yet.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3460 348d0f76-0448-11de-a6fe-93d51630548a
2010-05-28 18:27:25 +00:00
522dd7a5b2
Adding the variantrecalibration classes.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3459 348d0f76-0448-11de-a6fe-93d51630548a
2010-05-28 18:21:27 +00:00
2014837f8a
VariantOptimizer package is moved to core, renamed as VariantRecalibration, and added to the binary release package. VariantOptimizer walker is renamed to GenerateVariantClustersWalker and ApplyVariantClustersWalker renamed to VariantRecalibrator. Integration tests added, performance tests still to be done.
...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3458 348d0f76-0448-11de-a6fe-93d51630548a
2010-05-28 18:20:18 +00:00
rpoplin
kshakir
depristo
ebanks
hanna
delangel
aaron