Using Aroma.Affymetrix To Generate Cnv Calls

Question

[mducore]

I'm using the aroma.affymetix R package to do the segmentation of several normal genome Affymetrix 6.0 chip CEL files (arrays of log2 ratios and B allele frequencies). My goal is to generate absolute CNV (+/-) calls to do Fst calculations between populations. The problem is that the output I have from aroma gives me the segment data (start site, stop site, mean log 2 ratio, ect.), but not an absolute CNV state call. Does anyone know of an aroma command/tool that uses a statistical or algorithmic approach to generate the CNV calls? Or is there another package out there that could take the output from aroma.affymetrix and generate the calls from there?

thank you, -Michael

Answer 1

[bdemarest]

The easiest approach may be to use the 'canary' algorithm provided in Affy's Genotyping Console software (free!) That will give you categorical cnv genotypes for a set of known cnvs that are detectable with the SNP 6.0 platform.

You may also want to look into the CNVtools package for R. It performs sample clustering and assigns categorical cnv genotypes. The catch is that it works on one locus at a time. This paper (and especially the supplementary methods) may help explain cnv genotyping and why it is necessary: http://www.nature.com/nature/journal/v464/n7289/full/nature08979.html

Also, before using CNVtools, you need to determine common cnv locus break points for each cnv. I don't know of any standard software method for doing this with your own dataset. One approach is to use a list of known population cnv positions such as the DGV: http://projects.tcag.ca/variation/downloads/variation.hg19.v10.nov.2010.txt