I have access to the whole-genome sequencing data for 10 individuals (and a reference), half of whom are resistant to a certain treatment. As part of the effort to determine potential genetic causes of this resistance, I want to look at Copy Number Variation. I've come across a number of methods for identifying CNVs between a reference and a single individual (by read hybridization/alignment or read fragment length analysis), but I'd like to take advantage of the large dataset I have access to.
Can anyone point me in the direction of any statistical methods or programs which could compare two groups of multiple genomes, looking for read-depth differences?
Thanks!
-David
2 answers
CoNIFER is probably my favorite method for exon-capture CNV calling (although it would have to be modified for whole genome analysis, and I would typically use more than 10 samples):
http://conifer.sourceforge.net/
VarScan 2.0 provides CNV calls (I haven't tried it yet, but I like using VarScan for SNP calling):
http://varscan.sourceforge.net/
You might also want to look into the DNAcopy Bioconductor package:
http://www.bioconductor.org/packages/release/bioc/html/DNAcopy.html
GATK probably also provides some copy number functionality, but I haven't really been keeping track after it was announced that they would switch to a commercial license system.
I've tried using Pindel and BreakDancer for calling structural variants, but I haven't really liked the results that I have seen so far (and they aren't based upon read-depth differences, like CoNIFER and VarScan).
They don't solely work using read depth (they use discordant paired-read signals as well), but you might want to look into Genome STRiP:
http://www.broadinstitute.org/software/genomestrip/genome-strip
And CommonLAW:
Which are both designed to analyze multiple samples simultaneously.
Log in to answer this question.