mskcc/basicfiltering

Basic filtering scripts for various mutation calling tools

basicfiltering's Issues

Panel-of-normals (PoN) filter using VCFs listing non-reference events

• For each assay-type supported by Roslin, extract normal sample genotype columns from unfiltered VCFs, and merge into massive multi-sample VCFs per assay-type - listing all events observed in normal samples.
• Decide whether to create separate VCFs per variant caller.
• Remove normals that were reported as contaminated per Conpair.
• Remove normals with contamination from matched tumor by looking for supporting reads across known somatic hotspots.
• Develop thresholds or metrics to decide if a candidate somatic event is sufficiently common across the panel-of-normals i.e. they are either germline or a recurrent artifact.

Refer to related work done at /ifs/res/kandoth/panel_of_normals/readme to find Roslin normal BAMs enriched for recurrent artifacts, and the downstream normal_panel filter implemented here by @kpjonsson and last edited by @timosong. The primary benefit of using VCFs from variant callers is the ability to see complex indels as seem by the variant callers, and also not needing to genotype BAMs during the pipeline run. Stay in sync with @hellybelly330 who is doing similar work for the WES pipeline.