[jwag]

Jo,

The 1-50bp note is more about the default RTG mapping settings (you'd have to start adjusting the aligner settings to look for longer indels, at somewhat of a speed penalty). The variant caller will happily process anything you throw at it. The mappings in your original screenshot looked fine (in that there were reads spanning the deletion, plus some hanging into it, which all gets correctly taken into account by the caller during it's local realignment to the indel hypotheses, so I would say do *not* trim the ends of the reads after alignment).

Ok great, I'll give the RTG caller a go.

[jwag]

Jo,

The 1-50bp note is more about the default RTG mapping settings (you'd have to start adjusting the aligner settings to look for longer indels, at somewhat of a speed penalty). The variant caller will happily process anything you throw at it. The mappings in your original screenshot looked fine (in that there were reads spanning the deletion, plus some hanging into it, which all gets correctly taken into account by the caller during it's local realignment to the indel hypotheses, so I would say do *not* trim the ends of the reads after alignment).

Hi Len, I was able to run the RTG variant caller using the bam file produced using BBMap (without removing duplicate mapping reads and without trimming ends after mapping) and it seemed to work well. To calculate relative numbers of each haplotype from the VCF file, I took the allelic depths for the ref and alternate alles (AD) and divided it by total read depth at a variant location (DP). Using this, I estimated the rate of the large (~200bp) deletion at 25%, which is close to the numbers put out by other callers (19-34%).