Hi,

varscan mpileup2* strand-filter option states "Ignore variants with >90% support on one strand [1]" . When we have been running varscan with strand filter on, we have noted that varscan in some case still reports variants which fail the test.

examples:

chr17   41271293        .       GA      G       .       PASS    ADP=1025;WT=0;HET=1;HOM=0;NC=0  GT:GQ:SDP:DP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR    0/1:128:1025:1025:953:42:4.1%:1.4616E-13:35:29:947:6:**42:0** <- *100% support on one strand*

chr7  124499002       .       A       C       .       PASS    ADP=1561;WT=0;HET=1;HOM=0;NC=0  GT:GQ:SDP:DP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR    0/1:255:1561:1561:933:628:40.23%:7.9642E-222:35:0:845:88:**597:31** <- *95.1% support on one strand*

chr7    124499002       .       A       C       .       PASS    ADP=1779;WT=0;HET=1;HOM=0;NC=0  GT:GQ:SDP:DP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR    0/1:255:1779:1779:1055:724:40.7%:1.4608E-256:35:0:937:118:**690:34**  <-  *95.3% support on one strand*

The above examples indicate that there some other exception happening when variants are called. This exception allows these variants to bypass the strand-filter test.

Question: Has anyone come across this and worked out what this exception is? They clearly have >90% of allele support on one read.

Some of the biostar threads on this topic indicate that varscan calls alt alleles on the forward but this is not what we are seeing as a reason

When we switched this strand filter off this true positive comes through

chr13   32930609        .       C       T       .       PASS    ADP=1066;WT=0;HET=1;HOM=0;NC=0  GT:GQ:SDP:DP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR    0/1:255:1066:1066:573:488:45.78%:3.4741E-177:35:33:470:103:***440*:48**  <- *90.1% support on one strand*  and **we are seeing the alt allele on the forward** 

Question: The above example discounts the theory of forward allele calling by varscan or is there a further nuance to this?

Thanks in advance.