Hi all, I have been doing variant calling of somatic variants in exome seq of human genome using GATK pipeline based preprocessing and VarDict variant caller in paired end mode.
To my surprise, my colleague discoverred, that in VCF file the reported mutations do not correspond with BAM files. We inspected it by IGV and majority of mutations seem not to exist.
One option is that GATK preprocessing do bwa alignment to alternate locations and when the variants reported were to the alternate alignments IGV might not visualize them. Is there a way to extract alternate alignments from BAM files, e.g. by discarding primary alignments and keeping only alternate alignments to be able to sort bam with respect to them?
Thank you for suggestions.
Please also let me know if you have simmilar experience of caller reporting variants that you can not confirm by inspecting primary alignments in BAM files?
while it could be a bug , VarDict realign clipped reads so, you it could be a variant not visible in IGV unless you show the clipped bases.
I have shown clipped bases but the variants are still not there. I am not sure if IGV can show alternate mappings in BAMs since I guess there is no index for them. This is the only possibility I can think of, because when I later used mutect it reported filtering out variants in alternate alignments and the variants it reports can be indeed found in IGV.