Dan Koboldt

Thank you for your question. To me, it looks like you have your commands backwards: You run samtools mpileup FIRST, and pipe the output of that into VarScan. Also, keep in mind that you don't need to run two separate mpileup commands to generate the two-sample mpileup VarScan needs. You can give Samtools mpileup two BAM files. Also, a minor point, always provide an argument to VarScan option flags (e.g. use --mpileup 1). Here's my recommended restructuring of your command: samtools mpileup -B -q...

Hello, and thanks for asking. The short answer is no. The VarScan readcounts command has been around forever, and it does facilitate some specialized analyses. However, it does not replace the bam-readcount functionality required for false positive filtering (though this is high on the priority list for the next release).

Brian, If you extract Somatic calls only, you'll get only variants that are found in the tumor. You can use the processSomatic command to facilitate this extraction.

Brian, the ref column is the reference base. The alts correspond to alternate allele numbers 1 and 2, so you're correct that alt 1 is the alternate allele in the normal and alt 2 is the alternate allele in the tumor IN THIS CASE. The numbers representing alts are sample-agnostic, so it would also be possible for the normal to be 1/2. If there's not a comma-separated list, then the alt allele is the alternate seen, not necessarily in the tumor, but for this position. You have to look at the genotype...

1. BED-like var file for bamreadcount For SNVs it's straightforward as you say: chromosome, start, stop, and start equals stop. Indels are more challenging because of how they can be represented at different positions. I usually create a BED-like file for indels in which I print one line for every position from (start - indel_sizse - 1) to (stop + indel_size + 1). Bam-readcounts will return counts for all positions, and VarScan will use the best match to obtain metrics for filtering. 2. Running FPfilter...

Hello, thanks for your question. The input for VarScan fpfilter is NOT from VarScan readcounts. It instead requires a separate program, bam-readcount. This is a dependency we hope to fix with a future release, but if you use bam-readcount to generate the readcounts, you should be good to go.

Thank you for providing the examples; now I see what you were trying to describe. An ambiguity code in the reference is an unusual edge case, and a possibility that I had not considered (or coded for) in VarScan. It should be possible to add this compatibility for the next release.

Jose, That's an odd one! Would you mind providing a few lines of the relevant VCF output, and the mpileup for those positions?