What's the criterion of mapping quality when virmid was calling somatic variants?

[zeam]

I saw there is -q option that can be used to set "minimum mapping quality for sampling points". But what's the criterion of mapping quality when virmid was calling somatic variants? Because usually we want to use unique mapped reads to call variants (mapping quality > 0).

[Paul Wang]

hi,

I am not the author of Virmid, but I've been doing some in-depth comparison between different callers.

It seems like Virmid does consider all reads when calling for somatic variants. However, it has a site filter based on mapping qualities.

In the paper it explains this as:

If the mapping quality of the corresponding B allele read is significantly worse (>30 MAPQ score) than for A allele reads or if the overall ratio of ambiguously mapped reads (<17 MAPQ score) is above a threshold (>0.4).

However, in the code, the check performed is:

boolean rejectByMappingQual = matchqualMean - misqualMean > 10 || mq30ratio > 0.5 && matchqualMean - misqualMean > 5;

where:

matchqualMean = mean mapQ score of all reads in tumour sample

misqualMean = mean mapQ score of reads containing non-ref base at the site in tumour

mq30ratio = ratio of reads with mapQ<=30 to all reads in both normal and tumour samples