Gene based analysis methods, tools

Question

[zx8754]

Any suggestions on tool/methods for gene based analysis?

Data: I have all the GWAS data for case-control samples n=100K:

• genotyped GWAS (~600K SNPs, maybe useful for LD)
• imputed GWAS (IMPUTE Phase3, more SNPs)
• meta summary data based on multiple imputed GWAS data (IMPUTE2 Phase3, maybe more power, p-values)

Aim: X, Y, and Z genes are significant when all the SNPs within the gene boundaries (with flanks?) p-values are collapsed.

I remember testing VEGAS (5-6 years ago), thought now maybe new better tools are around.

Side question, what is the suggested boundary flanks for genes, 10K, 20K? These boundaries seems silly when genes are tiny.

I would prefer to stay in R world, but happy to switch if the tool is good enough.

Let me know if you need more info.

Answer 1

[Bioinformatics_NewComer]

Tools: GCTA and FAST (1.8). FAST later version is unstable with frequent crashes.

http://gcta.freeforums.net/thread/309/gcta-fastbat-based-association-analysis

For regions, it usually depends on the population and their LD pattern. You may go with 100KB on each side, or/and 250KB on each side.

For R, there's COMBAT package. It allows Vegas and one more gene-based test.