Upped the WGP VQSR memory to 32g to power through the filtering whole genome. TODO: Figure out what the right amount is.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5940 348d0f76-0448-11de-a6fe-93d51630548a
Reviewed pipelines with dev team.
HSP updates:
- Calling SNPs and Indels at the same time then using SelectVariants to separate them for filtering
- Moved logs next to the files like in WGP
- Flattened outputs into one directory
- The file names for the final outputs are now <projectName>.vcf and <projectName>.eval
- Updated test to pass the chr20 intervals instead of a boolean
- Removed MultiFCP
WGP updates:
- Only cleaning and calling chromosomes 1-22, X, Y, MT
- Splitting SNPs from indels, filtering indels, then merging the selected SNPs and selected Indels back together to make sure there are no collisions in CombineVariants
- Still running VQSR on the recombined SNPs plus hard filtered indels
- Using hard indel filters from delangel
- Reduced number of tranches with rpoplin
- Changed prior for dbsnp from 10 to 8 with rpoplin
- Assuming identical samples on both CombineVariants
- Explicitly using variant merge option UNION even though it's the default
- Not setting the default genotype merge option PRIORITIZE
- Generating a vcf and eval for each tranche
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5825 348d0f76-0448-11de-a6fe-93d51630548a
kshakir
depristo