delangel ca7810f11d First major update of indel genotyper: ... a) Really fix this time strand bias computation for indels, previous version was a partial fix only. b) Change way in which we deal with bad bases at the edge of reads. Even if a base is soft clipped in CIGAR string, there may still be dangling bases with Q=2 that may throw off QUAL computation in some sites. So, we're stricter and we also trim off those bases off read edges even if they are not soft-clipped officially. c) First feeble-minded attempt at runtime optimization - don't compute log and 10^base_qual every time. Rather, cache 10^-k/10 and log(1-10^-k/10) for all k <=60. This speeds up code about 4x. d) Further optimization: don't compute log(10^x+10^y) but rather use softMax function recently put into ExactAFCalculationModel. e) Skip bad reads where all Q=2 (sic) f) Avoid log to lin and back to log conversions of genotype likelihoods - this was legacy code from back when exact model did stuff in linear domain. This improves precision overall. git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4802 348d0f76-0448-11de-a6fe-93d51630548a		2010-12-07 18:35:22 +00:00
..
config	Provide a default logger, some config settings, and some doc updates.	2009-04-29 02:06:05 +00:00
src	First major update of indel genotyper:	2010-12-07 18:35:22 +00:00
test	BAQ calculation refactoring in the GATK. Single -baq argument can be NONE, CALCULATE_AS_NECESSARY, and RECALCULATE. Walkers can control bia the @BAQMode annotation how the BAQ calculation is applied. Can either be as a tag, by overwriting the qualities scores, or by only returning the baq-capped qualities scores. Additionally, walkers can be set up to have the BAQ applied to the incoming reads (ON_INPUT, the default), to output reads (ON_OUTPUT), or HANDLED_BY_WALKER, which means that calling into the BAQ system is the responsibility of the individual walker.	2010-12-06 20:55:52 +00:00