small case-control exome sequencing study

Question

[User6891]

Hi everyone,

I have exome data of 25 unrelated patients and 40 unrelated control samples. I'm looking for rare variants associated with the disease.

I've checked some obvious things: one variant present in cases not in controls. One gene enriched for rare variants in cases compared to controls.

However, now I want to do some statistical testing to find association of a variant/gene with the disease. Since this is just a small-scale study I don't think I can work with methods used in GWAS. Can someone point me to some papers/methods/ideas that could be of interest for my specific situation?

Thanks in advance

Answer 1

[Katie D'Aco]

As Sam pointed out, SKAT is a good place to start but unless there is a very strong signal from one gene/variant it's unlikely that you will get a statistically significant association from 25 cases. This paper from Lander & co has some calculations to estimate minimum sample size needed in these types of studies to get good results, and even under the best conditions there generally needs to be hundreds of cases. But it's worth a shot and even if nothing comes up with a low enough p-value the top hits still might be of interest.

If you haven't already, try answering questions such as: "what rare variants show up in at least two cases, but no controls", "what genes contain novel variants in at least two cases, but no controls"

Have you done any filtering by types of variants (eg, removing all synonymous variants from your analysis, or only looking at stop loss/gain variants)?

Answer 2

[Sam]

My suggestion will be trying to look into SKAT, our lab is trying to use that for the association test on exome sequencing. You might also try to perform annotation or prioritization using plinkSeq or kggSeq

If you are interested, you can always read into their paper