Entering edit mode
2.3 years ago
vpsev3
▴
20
I came across this site https://www.centogene.com/diagnostics/whole-genome-sequencing-with-centogenome.html
Are these low sensitivity rates related to an alignment problem?
I was wondering if, for example, PE150bp shotgun sequencing with 30X depth could detect these variants (like a 22q11.3 microduplication) with higher confidence.
I think you need to create one post "what can you detect with PE150bp shotgun sequencing with 30X depth " and not discussing each variant in a separate topics =)
I don't really understand what is large indel - 50 bp? Usually these ones are fine.
I don't see why they call some deletions as "copy-number deletions".
If you want to get good SV calling accuracy, check the long read alternatives to short-read WGS.
In our experience, with WGS you can get better sensitivity than this with large deletions at least, but it really depends on the analysis. @German.M.Demidov, maybe the distinction between deletions (for <10kb) and copy-number deletion (>10kb) is related to the software they use to identify them.
In any case, I think this low sensitivity might be due to the specificity guarantee (>99.9%), which will force them to be very stringent and only report very clear variants. There's always a tension between specificity and sensitivity and it seems they have decided to g all-in on specificity.
ah I see, thanks, however any deletion is a copy-number change of this particular piece - first time I see anyone using this terminology...Yeap, I assume that >10kb they detected with read depths and <10kb with paired-ends - maybe it was easier for them to call them differently...
Completely agree that any deletion is a copy-number change and that it does not make sense to call it otherwise.