search

walaj / svaba

Structural variation and indel detection by local assembly
GNU General Public License v3.0
231 stars 44 forks source link

Calls Specific to High Coverage Data #51

Open mej54 opened 5 years ago

mej54 commented 5 years ago

Hi,

We have been running multiple SV callers on high coverage data (275x tumor / 156x normal). We have observed that SvABA seems to be making quite a few SV calls that are about 500bp in size and only have a couple supporting reads. SvABA does not make these calls on the same tumor/normal pair with lower coverage (80x tumor / 40x normal). The other SV tools (Manta and Lumpy) do not find these calls at high coverage or low coverage, and after looking at these variants in IGV they do not seem real. We are wondering if this might have something to do with the reassembly being affected by high coverage data?

Thanks, Molly

walaj commented 5 years ago

Hi Molly,

Do you know if these are being reported as DSCRD events, or ASSMB events? That will change my approach to this. If DSCRD, there may be an issue with it not being able to accurately estimate the insert-size distribution. If ASSMB, there could be low-level sequencing artifacts that are a bit harder to get rid of (usually the way to fix is to post-filter the SV calls).