Question: Adding IDT CNV Backbone Spike-in to Small Targeted Panel for CNV Calling With CNVkit
gravatar for erikt
20 months ago by
erikt0 wrote:


I am designing an IDT capture panel consisting of 31 genes for sequencing some FFPE tumor samples. In addition to calling small variants I would also like to use CNVkit to call copy number aberrations for the genes in the panel. Since this is a small panel with only ~ 1300 probes it seems like it would beneficial to spike in IDT's CNV Backbone panel which would add ~ 9000 probes spaced about every .34Mb. But, since I am interested in low frequency variants in the 31 genes and the CNV backbone is considerably larger than the main panel, to spike the backbone in at an equi-molar ratio would significantly increase sequencing costs. Would it be feasible to spike the backbone in at say a 1:10 ratio, targeting about 100x average coverage? If so, does it make more sense to treat the backbone targets as part of the panel for CNVkit analysis, or let the program consider them off target reads?



cnv panel cnvkit targeted • 763 views
ADD COMMENTlink modified 20 months ago by markus.riester510 • written 20 months ago by erikt0
gravatar for markus.riester
20 months ago by
markus.riester510 wrote:

No experience with the IDT backbone, but yes, it makes sense to spike in the backbone lower. Usually you add single baits for the SNPs and multiple baits for each target in your 31 genes.

You want the backbone targets included since those are enriched in heterozygous SNPs and will thus help with LOH calling (or even segmentation when BAFs are used).

ADD COMMENTlink modified 20 months ago • written 20 months ago by markus.riester510

Thanks for your reply, Markus. I'll merge the two bed files in that case. My biggest concern is that a large difference in coverage between the two sets of targets would be a problem for the pipeline. Of course no capture has uniform coverage, and the pool of normals coverage profiles will exhibit the same coverage pattern.

ADD REPLYlink written 20 months ago by erikt0

Exactly. The PoN should normalize the coverage well. You might see a higher average coverage variance in the backbone due to the lower coverage (resulting in higher log-ratio noise), but with >100X this should be minor and most recent tools including CNVkit can incorporate this in the segmentation.

9000 probes is quite a lot. Which is nice of course, but if you are concerned about sequencing costs, and if you can subset the backbone, you could decrease the resolution to 1Mb in genomic areas where LOH calling is not as important.

Anyways, good luck with the panel design!

ADD REPLYlink written 20 months ago by markus.riester510
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 2126 users visited in the last hour