Regions of extremely high depth

[tcashby]

Hello, I was wondering if there was a way to have strelka skip regions of extremely high depth (e.g. a region with depth > 1000 in a WGS file of 40x mean coverage). I've got some samples that are waiting on one region to finish that have been running for more than 30 hours...

If not, is there any suggestions on how to move forward when I have this issue?

Thanks!

[ctsa]

@tcashby - Sorry for the delay, many of us have been out of office. Strelka is designed to curtail its realignment procedures based on absolute complexity thresholds as well as at very high depth relative to chromosome median, so if you have a high-depth region where the caller is getting 'stuck' this is definitely a bug. If this data can be shared, we'd be happy to take a look and fix it.

To skip a region, you could use the --callRegions option (https://github.com/Illumina/strelka/blob/master/docs/userGuide/README.md#call-regions), but note that this is designed to accept a bed file of regions to call rather than skip, so you would have to invert your skip-regions bed file. The configuration usage for this option is:

--callRegions=FILE  Optionally provide a bgzip-compressed/tabix-indexed
                    BED file containing the set of regions to call. No VCF
                    output will be provided outside of these regions. The
                    full genome will still be used to estimate statistics
                    from the input (such as expected depth per
                    chromosome). Only one BED file may be specified.
                    (default: call the entire genome)

[ztdy]

test.zip

[ztdy]

Please test the strelka2 on test.zip. Thank you very much!

[andrewrech]

I am having a similar issue on one interval on chr16 in high depth exome data on 5/85 WES. It seems multiple possible related issues are open?

I am using hg38 masked #55

@ctsa any follow up on this issue?