Delly with PON

[Robert A]

We have targeted sequencing data of ~400 independent tumor samples with no matching healthy tissue. However, we have ~15 samples of healthy tissue.

Is it possible to contrast the tumor samples against a "panel of normals" in the Delly pipeline, and would that actually make sense?
Thanks

[Tobias Rausch]

Delly was designed for WGS data and for targeted data you should only consider INFO/PRECISE variants. For these SVs, you can indeed use a panel of normal to filter out artifacts from the targeted approach. It's probably the easiest to do this at the end once you have called SVs in all samples using simple breakpoint overlaps with `bedtools intersect`, for instance.

--
You received this message because you are subscribed to the Google Groups "delly-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to delly-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/delly-users/72041111-0efb-410e-9451-1ce43eb3acd8n%40googlegroups.com.

[tr]

Alternatively, you can also follow the somatic workflow in the README with your 15 control samples.

[Robert A]

Thank you for the reply! I am thinking if I should still to the "Germline SV calling", with genotyping with the cancer samples as a background, and then compare it to the control:

Workflow:

A: SV calling on the BAM files of samples

A: merge all SV calls into unified site list
A: use merged calls list to genotype samples
A: merge all genotyped samples
a: use merged genotype list for germline filtering
B: SV calling on the BAM files of control
B: merge all control SV calls into unified site list

C: "setDIFF" genotyped sample SV calls with contol SV calls

Or can I leave out "a". I guess using the cancer samples as background is not necessary, if you have a control.

[Robert A]

I am not sure I am doing this right:

in the initial `delly call`, I get a few SVs. In one example only three. Then I merge, an do genotyping. Now, ALL the genotype .vcf's have 1233 SVs, most of them "precise" and with good quality. Where do they come from?

The Result is, that almost all samples have all possible SVs, which is definitely not correct.

I guess I am applying the genotype stop wrong