7

I'm currently attempting association analysis with an extremely small set of patient exomes (n=10), with no control or parental exomes available. Downloading the ExAC VCF of variant sites (http://exac.broadinstitute.org/downloads) or the 1000G integrated call sets (http://ftp.1000genomes.ebi.ac.uk/) and combining this with our pooled patient VCFs has not been successful (I suspect the approach of attempting to merge such large VCFs generated from different pipelines is rather naive).

Looking at the primary literature, I have gathered it should be possible to use these resources to help increase statistical power for our analysis. My question is how do I take these large .vcfs with many samples and successfully merge them to our patient .vcfs, such that the combined VCF can be used downstream to run analysis packages? (PODKAT, PLINK, etc.)

Daniel Standage
  • 5,080
  • 15
  • 50
carsweshau
  • 71
  • 2

1 Answers1

1

I gather that you want to use the background SNP frequency as a prior for input to your SNP calling algorithm?

I'm not sure of a canned algorithm for doing this, but a quick google shows up some promising links:

bli
  • 3,130
  • 2
  • 15
  • 36
Dan
  • 612
  • 3
  • 12
  • Thanks for the response - to clarify, I need additional details on how to correctly assign my priors using these public resources to jointly analyse multiple samples (our patient exomes and the population controls). The first link touches on this, but I'm unsure how to actually implement this approach.

    The third link would be useful, except our coverage is fine and instead we lack the sample size they have by several orders of magnitude (gulp)!

     – carsweshau Jun 22 '17 at 04:08
  • Yeah, I don't know of any tools that do this using existing data. What you /can/ do is re-analyze your data in the context of the other data in a single multi-sample call. This seems like a hack TBH, but should work. – Dan Jun 25 '17 at 21:55