We have a working NGS analysis variant calling pipeline for cancer which we've recently ran on a sample.
A certain mutation was found, specifically in chr3:178936095 (hg19), now as part of a confirmation process the lab ran a Sanger sequencing process on the mutation region and found out that a 2-based deletion and 1-base insertion wrt hg19, occurred in a close position, in chr3:178936116 and that the mutation we found seems like a false-positive.
Looking through the VCF files, I didn't find the deletion+insertion position, nor did I find it when I checked the BAM file.
- BWA MEM as an aligner. Our command for running is: bwa mem -t 12 -R <read groups info> <ref genome> <fastq files> > aligned.sam
- Realigning around indels using GATK.
- Calling variants with FreeBayes, GATK & bcf tools.
- Currently there's no filter based on read quailty other than the default one in bwa mem.
there was a consensus between all 3 on the mutation we found from the NGS data.
This has led me to think that the reads 'supporting' that hypothesis weren't aligned from the .fastq files as they were too different from the reference or got a 'lower score'.
My question is: Has something similar ever occurred to you? Is there a parameter/argument I can provide BWA MEM with so that those missing reads would be aligned? Otherwise, can you recommend a program/pipeline I can use to call these indels/insertions successfully?
Thank you very much in advance,