Has anyone tried this before?
We have data from the HumanExome-12v1-1 genotyping array from 16 individuals, 4 African Americans and 12 European Americans and we are hoping to find large CNVs using PennCNV. The HumanExome array contains many unique SNPs not present in the hhall.hg18.pfb, preventing us from simply extracting the PFBs for our SNPs (as in Pfb File For Cnv Analysis Using Illumina 610 Quad V1 Arrays).
Similar analysis has been done with a large cohort by calculating the population B allele frequencies from the samples, but our small sample size make me wary of using this approach. (http://www.ncbi.nlm.nih.gov/pubmed/23938935)
To get around this, we are planning to calculate the PFBs directly from 1000 genomes VCFs. I can foresee some problems with this approach but it might be worth a try.
Hi, what did you end up doing? Although I have a large sample size, I have many related (and non-white) individuals, so I also think 1000 genomes might be a good source.
All the best,
Steve
Hi Steve,
We ended up doing this and the approach seemed to work technically. However, the results were not very interesting so we did not put too much time into these data.