I have a significant SNP from a Genome Wide Association Study and I want to find correlation between change in this SNP between samples and change in gene expression.
I've manually used ANOVA in R which shows that many genes expression are associated with this SNP.
I'm wondering if there is another tool that anyone has used to perform this kind of analysis that would be staightforward to use. "Interval Mapping" seems to be the more accepted method for eQTL finding.
I've been trying to do this in GGtools but find the documentation to be limited which doesn't make life too easy.
Karl Broman (author of the very highly recommended R/QTL library) maintains a nice list of other software packages for QTL mapping on his RQTL web site. However, all of the software I'm familiar with is designed to do linkage analysis mapping loci in experimental genetic crosses (e.g. backcrosses or intercrosses with mice). When I have to map eQTL in human data, I've had to write my own scripts as you did using point mapping rather than the more informative interval mapping. A quick check of the Genetics view of bioconductor doesn't show anything obvious; there's a software gap here as far as I can tell.
The good news is that since you're a priori already interested in an exact locus, you have probably have already found the peak or near-peak of any interesting expression association with that locus in your dataset. The bad news is that you have a multiple testing problem; be sure to consider that when interpreting your results. It only gets worse if you consider all loci and all genes, though.
You can use SCANDB for scanning the SNP/regions for pre-computed eQTL. See the SNP and region tabs. I find it useful for a similar eQTL analysis. SCANDB manuscript is available here
You can use PLINK (http://pngu.mgh.harvard.edu/~purcell/plink/) for performing association on quantitative traits. The software is quite powerful and straightforward. You just need to add expression as a quantitative phenotype in your input file.
But does PLINK actually perform interval mapping, which is what Paul is asking about? I'm not a PLINK expert but I think it only performs point estimation.
But does PLINK actually perform interval mapping, which is what Paul is asking about? I'm not a PLINK expert but I think it only performs point estimation.