Calling tumor CNVs without normal as control?

[ymc]

Also curious about this. Perhaps they are using pooled normals?
If library prep & sequencer are the same, and the pooled samples are copy-number neutral, it should be possible to get decent results using ExomeCNV.

If I understand correctly, pooled normal only helps for calling CNVs on normal samples in the context of ExomeCNV.

I presume they aren't very nosy about accuracy. As long as it works in their clinical trials, then it is ok. One error can bite their ass though....

[bw.]

Also curious about this. Perhaps they are using pooled normals?
If library prep & sequencer are the same, and the pooled samples are copy-number neutral, it should be possible to get decent results using ExomeCNV.

[ymc]

http://www.foundationone.com/about-foundationone/ONE-1-002-20130904_FoundationOne_Technical.pdf

I noticed that the recently IPOed Foundation Medicine only used tumor sample to find mutations, CNVs and rearrangements by sequencing 200+ genes panel.

I think their approach is fine with mutations/indels but sequencing normal as well to differentiate the true somatic ones might be slightly better. For rearrangement/fusion genes, they are using some introns sequence for that. I think RNA-seq can cover more fusion cases than this approach, is that right?

The only thing that bugs me a lot is their claim about CNVs. Isn't that without the normal control, GC bias can't be removed?