Guillermo del Angel 20d3137928 Fix for indel calling with UG in presence of reduced reads: When a read is long enough so that there's no reference context available, the reads gets clipped so that it falls again within the reference context range. However, the clipping is incorrect, as it makes the read end precisely at the end of the reference context coordinates. This might lead to a case where a read might span beyond the haplotype if one of the candidate haplotypes is shorter than the reference context (As in the case e.g. with deletions). In this case, the HMM will not work properly and the likelihood will be bad, since "insertions" at end of reads when haplotype is done will be penalized and likelihood will be much lower than it should. ... -- Added check to see if read spans beyond reference window MINUS padding and event length. This guarantees that read will always be contained in haplotype. -- Changed md5's that happen when long reads from old 454 data have their likelihoods changed because of the extra base clipping.		2013-04-29 19:33:02 -04:00
..
annotator	Add support for snpEff "GATK compatibility mode" (-o gatk)	2013-04-26 15:47:15 -04:00
beagle	Updated all JAVA file licenses accordingly	2013-01-10 17:06:41 -05:00
bqsr	Trivial BQSR bug fixes and improvement	2013-04-11 17:08:35 -04:00
compression/reducereads	Various bug fixes for recent Reduce Reads additions plus solution implemented for low MQ reads.	2013-04-24 18:18:50 -04:00
diagnostics	Split class names into stratification and metrics	2013-04-24 14:15:49 -04:00
diffengine	Fixed issues raised by Appistry QA (mostly small fixes, corrections & clarifications to GATKDocs)	2013-03-12 10:57:14 -04:00
fasta	Updated all JAVA file licenses accordingly	2013-01-10 17:06:41 -05:00
filters	Don't allow users to specify keys and IDs that contain angle brackets or equals signs (not allowed in VCF spec).	2013-04-05 00:52:32 -04:00
genotyper	Fix for indel calling with UG in presence of reduced reads: When a read is long enough so that there's no reference context available, the reads gets clipped so that it falls again within the reference context range. However, the clipping is incorrect, as it makes the read end precisely at the end of the reference context coordinates. This might lead to a case where a read might span beyond the haplotype if one of the candidate haplotypes is shorter than the reference context (As in the case e.g. with deletions). In this case, the HMM will not work properly and the likelihood will be bad, since "insertions" at end of reads when haplotype is done will be penalized and likelihood will be much lower than it should.	2013-04-29 19:33:02 -04:00
haplotypecaller	Bugfix for ReadClipper with ReducedReads	2013-04-29 11:12:09 -04:00
indels	Fixed IndelRealigner reference length bug (GSA-774)	2013-02-19 16:00:36 -05:00
phasing	Updated all JAVA file licenses accordingly	2013-01-10 17:06:41 -05:00
validation	MathUtils.randomSubset() now uses Collections.shuffle() (indirectly, through the other methods	2013-03-29 14:52:10 -04:00
varianteval	Move some VCF/VariantContext methods back to the GATK based on feedback	2013-01-29 16:56:55 -05:00
variantrecalibration	Update MD5s for VQSR header change	2013-04-16 11:45:45 -04:00
variantutils	Fixed bug reported on the forum where using the --exclude_sample_file argument in SV was giving bad results.	2013-04-26 12:23:11 -04:00