Calling CNVs in exomes - CLAMMS

Question

[kath.a.fawcett]

Hello,

I would like to identify CNVs from a very large exome sequencing dataset (tens of thousands) and the CLAMMS software looks very promising (https://github.com/rgcgithub/clamms). Unfortunately, issues raised on GitHub are not responded to and I was wondering whether anyone out there had ever used this software and could help with a naïve question:

The CLAMMS documentation states: "CLAMMS collects seven QC metrics for each sample and performs a fast k-nearest neighbors search algorithm". However, as far as I can tell CLAMMS does not do this automatically as part of the model building or CNV calling steps. Does the user have to generate these QC metrics (using Picard) and identify nearest neighbours themselves based on the example protocol provided?

Thanks very much!

Kath

[lakhujanivijay]

Hi kath.a.fawcett

Unfortunately, issues raised on GitHub are not responded

yes, I have also raised several questions in the recent past and none of them were answered. So, IMHO, you should not expect any response from the author about this. I can confirm that CLAMMS does not generate QC as far as the model building and CNV calling part. I can say this confidently because I already have tried that.

Unfortunately, on a separate note, there is something wrong with my reference file because of which I am getting issues.

Answer 1

[lakhujanivijay]

The supplementary information here says

Query our laboratory information management system to retrieve seven Picard sequencing quality control metrics for each sample: GCDROPOUT, ATDROPOUT, MEANINSERTSIZE, ONBAITVSSELECTED, PCTPFUQREADS, PCTTARGETBASES10X, and PCTTARGETBASES50X

That means, they assume that we calculate that ourselves.