What's the best CNV tool for areas >1MBp

Question

[sebastiz]

I want to be able to detect copy number variants for amplifications and deletions >1MBp. I am told that FREEC is unreliable but I suspect that it's OK for detecting CNV of this size. I'm not really interested in smaller CNV sizes. Does anyone have any experience in the software for the detection of CNV of this size and larger and which might be the best to use or if it makes any difference for this- any references would be great.

Thanks

[iraun]

I've never use it but how about CNVnator? (http://sv.gersteinlab.org/)

Here there is a recent publication summarizing the computational tools for CNV detection: http://www.biomedcentral.com/1471-2105/14/S11/S1. Might be of interest to take a look.

Answer 1

[Chris Fields]

As airan mentioned, CNVnator is useful for some genomes (it was mainly designed for human/mouse IIRC). However, the genome you are aligning against, and in particular the difference in your sample vs that reference and how much of the genome is repetitive, makes a huge difference in what you should run.

For instance, CNVnator gave fairly terrible results on a plant genome we did recently, for both of the reasons above. The genome was ~80% repetitive and quite different than the reference. We ended up essentially getting much more information, and much less noise, by simply generating read coverage stats from genic regions and assessing those directly for gain/loss.