a) Modified the way clipped bases are dealt with in ReadPosRankSumTest when annotating indels. Cigar string cannot be trusted because BWA can clip good high quality bases and some sites get incorrect ReadPos annotations if BWA systematically clips at an indel breakpoint.
b) PL header needs to specify "." as length. Otherwise we fail VCF validation if multiallelic sites are present.
describe" as an absolute version number (if the repository has at least one
tag), using the raw SHA-1 hash value as a fallback version number in the case
where there are no tags.
1) There is now a different parameter for sample name (-sn), sample file (-sf) or sample expression (-se). The unexpected behavior of the previous implementation was way too tricky to leave unchecked. (if you had a file or directory named after a sample name, SV wouldn't work)
1b) Fixed a TODO added by Eric -- now the output vcf always has the samples sorted alphabetically regardless of input (this came as a byproduct of the implementation of 1)
2) Discordance and Concordance now work in combination with all other parameters.
3) Discordance now follows Guillermo's suggestion where the discordance track is your VCF and the variant track is the one you are comparing to. I have updated the example in the wiki to reflect this change in interpretation.
4) If you DON'T provide any samples (-sn, -se or -sf), SelectVariants works with all samples from the VCF and ignores sample/genotype information when doing concordance or discordance. That is, it will report every "missing line" or "concordant line" in the two vcfs, regardless of sample or genotype information.
5) When samples are provided (-sn, -se or -sf) discordance and concordance will go down to the genotypes to determine whether or not you have a discordance/concordance event. In this case, a concordance happens only when the two VCFs display the same sample/genotype information for that locus, and discordance happens when the disc track is missing the line or has a different genotype information for that sample.
6) When dealing with multiple samples, concordance only happens if ALL your samples agree, and discordance happens if AT LEAST ONE of your samples disagree.
---
Integration tests:
1) Discordance and concordance test added
2) All other tests updated to comply with the new 'sorted output' format and different inputs for samples.
---
Methods for handling sample expressions and files with list of samples were added to SampleUtils. I recommend *NOT USING* the old getSamplesFromCommandLineInput as this mixing of sample names with expressions and files creates a rogue error that can be challenging to catch.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@6072 348d0f76-0448-11de-a6fe-93d51630548a
Updated the Data Processing Pipeline to use the new Picard classes and reorganized the pre-processing of the pipeline accordingly.
Will only update the wiki once this change goes live.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@6071 348d0f76-0448-11de-a6fe-93d51630548a
Eric Banks
Mauricio Carneiro
Mark A. DePristo
Ryan Poplin
David Roazen
Matt Hanna
Guillermo del Angel
carneiro
ebanks
droazen