Hi. I am doing a matched Normal-Tumor mutation detection with whole exome sequencing data.
When evaluating my pipelines, I noticed that some somatic mutations are not showing if I follow the best practice for somatic detection. After I checked the bam files aligned by different pipelines with IGV, I found that the difference was made in the IndelRealigner process.
My question is, I have an impression that it is better to align the normal / tumor reads as input together, and then produce separated output through the --nWayout option in GATK IndelRealigner, but how would it be better to do so than aligning the normal / tumor sample separately? Comparing the separately-aligned result (1way) and aligned-together result (2way), I found a mutation showing in 1way output (with allele frequency 24/193), while it is almost gone in 2way output (1/170). I checked the alignment with IGV, and I think the mutation might be resulted by artifact, which means the the IndelRealigner made the right decision. However, I am still not sure if it is always better to trust the 2way result. I am worried about missing some somatic mutations and not knowing about it, since somatic mutation usually has lower allele frequency.
May I ask how to interpret the benefit of using 2way output over 1way?
Any reply would be greatly appreciated.