528f6344afMoved ReadBackedPhasingWalker to phasing sub-directory
fromer
2010-09-13 19:36:41 +0000
fa3be2209fImprovements to the error display code to print out the SVN number in all messages. Fixes to CallableLoci and tests to check for that case
depristo
2010-09-13 18:36:45 +0000
81c82ce134Fix for Queue
depristo
2010-09-12 15:18:08 +0000
7880863eb7Final step in error refactoring. GATK exception is now ReviewedStingException, indicating that this exception is really what one wants. Only use this exception when you have thought about StingException vs. UserException and made a real decision.
depristo
2010-09-12 15:07:38 +0000
7ad8fbdd5aMoved GATKException to exceptions
depristo
2010-09-12 14:47:19 +0000
40e6179911Penultimate step in exception system overhaul. UserError is now UserException. This class should be used for all communication with the USER for problems with their inputs. Engine now validates sequence dictionaries for compatibility, detecting not only lack of overlap but now inconsistent headers (b36 ref with v37 BAM, for example) as well as ref / bam order inconsistency. New -U option to allow users to tolerate dangerous seq dict issues. WalkerTest system now supports testing for exceptions (see email and wiki for docs). Tests for vcf and bam vs. ref incompatibility. Waiting on Tribble seq dict improvements to detect b36 VCF with b37 ref (currently cannot tell this is wrong.
depristo
2010-09-12 14:02:43 +0000
da2e879bbcMiscellaneous improvements to indel genotyper: - Add a simple calculation model for Pr(R|H) that doesn't rely on Dindel's HMM model. MUCH faster, at a cost of slightly worse performance since we're more sensitive to bad reads coming from sequencing artifacts (add -simple to command line to activate). - Add debug option to calculation model so that we can optionally output useful info on current read being evaluated. (add -debugout to commandline). - Small performance improvement: instead of evaluating haplotype to the right of indel (just with a 5 base addition to the left), it seems better to center the indel and to add context evenly around event.
delangel
2010-09-12 13:50:28 +0000
61d511f601Small memory performance improvement: remove the mapping from the hash instead of setting the value to null (i.e. remove the key too)
ebanks
2010-09-12 05:19:09 +0000
a0231f073fDamnit. Enabling the Picard code to recalculate all of the relevant SAMRecord attribute tags means that I need to have reference bases over all read bases even after realignment (and there are some big indels in dbsnp). Fortunately, I have my trusty IndexedFastaSequenceFile reader handy! Re-enabling the previously broken performance test.
ebanks
2010-09-12 05:06:37 +0000
87aca64716Jumped the gun a bit on bam on-the-fly indexing -- Tim says it's not ready yet. Turned it off by default and added a property to turn it back on.
hanna
2010-09-10 21:16:03 +0000
7b113a4886Truncate 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
rpoplin
2010-09-10 18:37:49 +0000
8f1a32acaeAll exceptions thrown by the GATK have been reviewed and UserErrors replaced where appropriate. Shazam. Another check-in will remove the GATKException and restore the StingException.
depristo
2010-09-10 15:25:30 +0000
cf33614ddcremove the test that's failing the performance tests, please don't release until this is figured out
aaron
2010-09-10 06:30:40 +0000
bd4bc84abdcomment out the broken aligner test again - I'll take a crack at fixing it tomorrow. Each software engineer is going to take a pass at fixing it, and we'll see who can do it with the most style.
aaron
2010-09-10 05:22:24 +0000
61e848c4f0It's clear from Sendu's calling and my own calling that -qScale 100.0 is a much better default value for low pass data.
rpoplin
2010-09-10 01:47:21 +0000
e183b6598c- Fix our private repository of bwa reference support files. - Update the test to point to our repository. - Update the md5 to reflect new Picard tag ordering.
hanna
2010-09-10 00:29:26 +0000
1de713f354Massive review of maybe 50% of the exceptions in the GATK. GATKException is a tmp. tracker so that I can tell which StingExceptions I've reviewed. Please don't use it. If you are working on new code and are considering throwing exceptions, it's either UserError or StingException, please
depristo
2010-09-09 23:21:17 +0000
dfdd0b69a9Removed unused dependency (it was causing a problem by looking for an X11 connection that didn't necessarily exist).
kiran
2010-09-09 19:56:00 +0000
f5c295b6b2add a little bit of documentation to the RMD track builder and wrap any exceptions thrown in tribble with the file source and line that caused the error.
aaron
2010-09-09 17:56:36 +0000
aeb897db7fVR walkers look at by-hapmap validation status by default. Eric will be updating the syntax to allow for more flexibility here.
rpoplin
2010-09-09 15:40:56 +0000
d7f55574e2Re-enabling aligner integration test now that we're back to having more than 1 or 2GB memory.
kshakir
2010-09-09 15:09:48 +0000
d625186796I think the VR integration tests are fine.
rpoplin
2010-09-09 15:00:41 +0000
6a30617a60Initial 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.
depristo
2010-09-09 11:32:20 +0000
8708753a6acheckin for removal
depristo
2010-09-09 02:32:46 +0000
5119bdb55e- Update DoC to support output to /dev/null. - Add a release sanity check for DoC. - Update release sanity checks with new command-line argument system.
hanna
2010-09-08 23:43:18 +0000
6c27a9bfc6- Include the fasta index builder in the package. - Unit/integration test against the jars built to the dist directory, so that the tests reveal classes that don't exist in the final jars.
hanna
2010-09-08 20:37:54 +0000
ce031b2f05PhasingEvaluator prints out interesting sites (only 1 phased, or phases disagree)
fromer
2010-09-08 18:21:21 +0000
40283f6456Success! TranscriptToGenomicInfo now works without the delicate hacks that Ben had put in.
ebanks
2010-09-08 18:06:00 +0000
cd091d7309This walker can NOT be tree-reducible (in its current state). Given that it's meant to be run just once for any given transcript set, this is not at all a problem.
ebanks
2010-09-08 16:47:51 +0000
ae9cba1c73After an epic battle with this code until 3am last night, I have discovered that it is tragically and fatally busted. Ben clearly didn't understand how the ROD system works when writing it and so it is unusable in its current state. I've ripped out all code and it now gracefully exits telling the user that we are actively working on a replacement for this tool. Sigh.
ebanks
2010-09-08 16:39:41 +0000
cd2bfb09efChange for Tim: invalidate the MD tag (temporarily) if it exists in a read that gets realigned
ebanks
2010-09-08 13:59:09 +0000
65edbced36Addition for Tim: recalculate the NM and UQ tags after realignment. Also, don't fix the insert size calculation, since that's done by fix mate information.
ebanks
2010-09-08 04:02:14 +0000
84ddadca64One more fix: exclude input VCF file's directory name from the output file.
kiran
2010-09-08 03:09:08 +0000
594fb4a547More plots in report
depristo
2010-09-08 02:56:51 +0000
c524a9ec83Minor usability improvements to reporting tool
depristo
2010-09-08 02:56:25 +0000
71046e650eAdded a more robust check for Jishu -- am pretty sure the .bam header is busticated
chartl
2010-09-08 01:11:22 +0000
ae3f7026a4Corrected phasing quality evaluation to correctly account for hom sites that break phase
fromer
2010-09-07 22:43:54 +0000
bd878565f6Change for Firehose: infers output path from the input VCF, so that we don't have to change a whole bunch of stuff so that Firehose knows where to expect the output file.
kiran
2010-09-07 21:07:46 +0000
501f6a0e14Temporary hack to disable index creation when target BAM is /dev/null. Tim promises me that Picard will put in a real solution next week.
hanna
2010-09-07 16:57:51 +0000
5d0d9c7dceMy parallel version of TranscriptToInfo now emits 'chr start end' instead of 'chr:start-end' for records so that 1) they can be easily sorted in coordinate order (allowing me to emit records out of order if I choose) and 2) the file can be tabix indexed (when we stop finding 'critical' bugs in that code).
ebanks
2010-09-07 05:20:40 +0000
fe2a0bb3a6Fix for DoC issue with multiplexer -- will retire use of multiplexer when GATK reporting structure comes online.
hanna
2010-09-07 00:44:07 +0000
19e22cfa87Fixed a bug where the script looked for the wrong column name. Also, all results are now returned in a single plot.
kiran
2010-09-06 14:19:57 +0000
4d4ef5b42cIn the end, it's not worth rewriting TranscriptToInfo from scratch. I'm keeping the old one around for a bit so I can play with this new version which 1. doesn't store the records in memory so can be run in under 1Gb of memory, 2. actually emits all of the records (the original fails in some cases), and 3. is refactored to cut out ~20% of the code.
ebanks
2010-09-06 02:37:34 +0000
0dd5a0990dNow annotates sites marked as filtered out (this is important if sites are in a lower-quality tranche).
kiran
2010-09-04 00:36:55 +0000
e9af893bf4Write headers that are VCF4.0 compliant
kiran
2010-09-03 23:46:27 +0000
ef7454a241Minor improvements to indel genotyper: a) Ability to specify haplotype size from command line b) Expand reference context window so we can form haplotypes for longer indel events. c) small bug fix in temp output writer (to be removed once I can emit vcfs)
delangel
2010-09-03 22:52:08 +0000
7eeabe534aQSample walker for 1KG -- measures aggregate quality of sequencing. Includes misc. improvements throughtout the code, including using the new Tribble GenotypeLikelihoods class for working with VCF GLs from the UG
depristo
2010-09-03 18:21:43 +0000
e3962c0d13VR integration tests are longer but much more useful.
rpoplin
2010-09-03 15:50:19 +0000
da11efa1a2Automatically write BAM file indices for coordinate-sorted BAMs.
hanna
2010-09-03 14:10:44 +0000
60003aeacaBug fix in VariantRecalibrator. Only add sample names from the input rod bindings, not from all rod bindings.
rpoplin
2010-09-03 13:31:49 +0000
3c9597d45aOnTraversalDone writes output to out now
depristo
2010-09-03 12:55:03 +0000
73d41bfa24CountLoci nows writes out to a file for Queue status tracking. VariantAnnotatorEngine has a special group None that doesn't add any annotations; useful for those who are testing UG performance
depristo
2010-09-03 12:52:33 +0000
b59d62927eFix busted performance test (-outputBam has been deprecated in the BQ recalibrator in favor of -o)
ebanks
2010-09-03 12:51:53 +0000
70bb480939The battle is over. Picard is revved.
hanna
2010-09-03 05:28:01 +0000
fdaac4aa78As the VCF guru, I'll take this one for Andrey. Someone has actually found a deletion at the beginning of the chromosome. Instead of failing with an ArrayIndexOutOfBoundsException, just don't try to print out the record. Our VCF writer doesn't really support this case (yet).
ebanks
2010-09-03 03:27:43 +0000
c45ffcdaedChanging documentation (temporarily) to warn people that -U is not supported.
ebanks
2010-09-03 03:18:07 +0000
8a7f5aba4bFirst more or less sort of functional framework for statistical Indel error caller. Current implementation computes Pr(read|haplotype) based on Dindel's error model. A simple walker that takes an existing vcf, generates haplotypes around calls and computes genotype likelihoods is used to test this as first example. No attempt yet to use prior information on indel AF, nor to use multi-sample caller abilities.
delangel
2010-09-03 00:25:34 +0000
a1cf3398a5Added basic version of phasing evaluation: GenotypePhasingEvaluator
fromer
2010-09-02 22:09:50 +0000
fd5970fdd4At chartl's superb suggestion, command line files are now all Files instead of old method of sometimes "has a File". Should be easier when reassigning them. No longer generating deprecated GATK arguments on the Queue extensions. Emitting deprecation warnings to Queue compile to help debugging issues.
kshakir
2010-09-02 21:30:48 +0000
0bb05fb472Bug fix in VariantRecalibrator. Only add sample names from the input rod bindings, not from all rod bindings.
rpoplin
2010-09-02 21:12:09 +0000
3a4844ebdeAdditional partition types into DepthOfCoverage:
chartl
2010-09-02 19:30:03 +0000
590bb50d16Test for missing read group
chartl
2010-09-02 14:22:13 +0000
acd6bd2430Experimental tool to annotates indels that are provided in a VCF file based on RefGene. Specifies gene, transcript, strand, type (Non-frameshift, frameshift, 5'-UTR, 3'-UTR, SpliceSiteDisruption, Intron, or Unknown).
kiran
2010-09-01 23:30:28 +0000
dc5f858d29Replaced placeholder support for splitting by read group with read support (sorry everyone), and added relatively comprehensive unit tests to ensure that splitting by read group works.
hanna
2010-09-01 22:24:50 +0000
b28f63a948Base recalibrator now uses -o and deprecates -outputBam
rpoplin
2010-09-01 22:13:50 +0000
33400074faUpdated tribble BED parsing code to use the official UCSC spec, and updated tests to match expected results.
kshakir
2010-09-01 21:49:06 +0000
ca503e5801Queue scripts for recalibration and running nSample UG jobs pre and dynamic merging
depristo
2010-09-01 20:23:37 +0000
995cfe34feYou can have an error so early that some engine fields are uninitialized. Commit protects RunReport from these errors
depristo
2010-09-01 19:00:25 +0000
a975db2c2eBug fix for the case of reads with no read bases!
rpoplin
2010-09-01 16:58:54 +0000
0c54bf4195Better reporting and now with a special mode for listing exceptions
depristo
2010-09-01 16:19:51 +0000
cdad243645updated version of the DPR. Now produces part of the tearsheet as well as good depth of coverage figures
corin
2010-09-01 15:38:58 +0000
469bbaa240Added more integration tests for the variant quality score recalibrator
rpoplin
2010-09-01 15:31:24 +0000
fc5caa98a5Improved reporting now with metrics by day/week/etc.
depristo
2010-09-01 02:43:13 +0000
96040726acBetter exception text for the common error of providing only dbsnp but giving dbsnp sites zero clustering weight.
rpoplin
2010-08-31 21:36:43 +0000
8683087756Suppl. tools for working with and displaying GATK run reports
depristo
2010-08-31 20:32:22 +0000
32c6b48106Proper memory metrics in the file. Please use -et if at all possible
depristo
2010-08-31 20:30:09 +0000
63c7cbd89bForgot to commit this long ago, change so the tables are correctly propagated
chartl
2010-08-31 19:06:52 +0000
db4ff7317fallowing empty RMD files (we need to not validate their sequence dictionaries against the reference in this case)
aaron
2010-08-31 17:45:33 +0000
3d6c4fc55fRemoving the obsolete --hapmap and --hapmap_chip options
ebanks
2010-08-31 16:57:05 +0000
fromer
depristo
delangel
ebanks
hanna
rpoplin
aaron
kshakir
kiran
chartl
corin