8ac6d3521f
-- AssessNA12878 now breaks out multi-allelics into bi-allelic components. This means that we can properly assess multi-allelic calls against the bi-allelic KB -- Refactor AssessNA12878, moving into assess package in KB. Split out previously private classes in the walker itself into separate classes. Added real docs for all of the classes. -- Vastly expand (from 0) unit tests for NA12878 assessments -- Allow sites only VCs to be evaluated by Assessor -- Move utility for creating simple VCs from a list of string alleles from GATKVariantContextUtilsUnitTest to GATKVariantContextUtils -- Assessor bugfix for discordant records at a site. Previous version didn't handle properly the case where one had a non-matching call in the callset w.r.t. the KB, so that the KB element was eaten during the analysis. Fixed. UnitTested -- See GSA-781 -- Handle multi-allelic variants in KB for more information -- Bugfix for missing site counting in AssessNA12878. Previous version would count N misses for every missed value at a site. Not that this has much impact but it's worth fixing -- UnitTests for BadSitesWriter -- UnitTests for filtered and filtering sites in the Assessor -- Cleanup end report generation code (simply the code). Note that instead of "indel" the new code will print out "INDELS" -- Assessor DoC calculations now us LIBS and RBPs for the depth calculation. The previous version was broken for reduced reads. Added unit test that reads a complex reduced read example and matches the DoC of this BAM with the output of the GATK DoC tool here. -- Added convenience constructor for LIBS using just SAMFileReader and an iterator. It's now easy to create a LIBS from a BAM at a locus. Added advanceToLocus function that moves the LIBS to a specific position. UnitTested via the assessor (which isn't ideal, but is a proper test) |
||
|---|---|---|
| .. | ||
| R | ||
| c | ||
| chainFiles | ||
| doc | ||
| java | ||
| keys | ||
| packages | ||
| perl | ||
| scala | ||
| testdata | ||