I work on mutation detection of paired end illumina NGS data, for this I use Varscan and Lofreq, but I have some isseus with strand bias, I hoped I could use the dp4 counts and the ADR and ADF counts but those count the forward and reverse reads instead of the forward and reverse original strands - based on the read pair orientation.
To illustrate an IGV screenshot below, reads are colored by first of pair strand - the forward strand is red, and reverse is blue There are 12 reads with a mutation G > T - 6 forward and 6 reverse - but those are all on the forward strand - So if I want to filter on strandbias the counts of Lofreq and Varscan are useless.
Does somebody know a solution?
You may have to explain the problem a bit more. Be aware that IGV filters your reads, and so do [I presume] VarScan and LoFreq. It is next to impossible to get all of these programs to agree on things.