Hi all, We are working on some CtDNA data right now, and still lacking good software for SV calling. Please share you thoughts. Thanks,
Circulating tumor DNA (ctDNA) is generally extremely heavily fragmented. Single-end sequencing/short fragments have essentially no read pair SV signal and any soft clip signal will be weak. Additionally, for any given variant, it is not a simple problem to assign it as a tumour or germline event since you also lack your traditional tumour/normal sample pairings.
The lack of good software for SV software for ctDNA is likely due the above reasons. You can use generic SV calling software (such as my tool GRIDSS) for calling ctDNA SV events but be aware that you are likely to have relatively low sensitivity and a high false discovery rate.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Thank you for the suggestion. Can you recommend settings to use with GRIDSS on CtDNA? Data we have is illumina pair end. Samples from white blood cells with similar read depth are used as control to achieve somatic sample pairing.
The default settings should be fine. Are you using doing pair-end sequencing, or just single end? If you just have single end reads, then GRIDSS is unlikely to do any better than a purely split read based caller (e.g. CREST). Due the issues with ctDNA I have outlined above, you may find that it might not be possible to make "good" SV calls from your data.
We had same problem. We have paired-end targeted sequencing and insert size around 200 bp. Any suggestions are highly welcome.