Snp Association Next Gen Sequencing

Question

[Travis]

Hi,

If I have targeted gene resequencing data (1000 genes) for 75 individuals in group A and 75 in group B, is there an established method of looking for SNPs or Indels associated with the two groups?

For example, I know I could align to a reference genome with Bowtie and then call SNPs for each individual with SAMtools, but how would I then go about considering the 75 Group A vs Group B?

Is there anything that allows this kind of analysis or would I have to design scripts myself?

Thanks in advance.

Answer 1

[David Quigley]

Once you have genotypes, you are doing an association study. Presumably you are interested in whether a particular allele is over-represented in one group compared to what you would expect by chance. If you don't want to use R and write the code yourself, the PLINK package is very mature and easy to use. There are various tests you could consider; you might look first at the allele test and Cochran-Armitage. Note that unless you have an incredibly powerful genetic effect, you are likely to be very underpowered to detect anything with 75 cases/controls. If you're just getting into this kind of study you should look into some methodology reviews (e.g. check out the Nature Reviews: Genetics series on GWAS) as there are a lot of subtle things that can go wrong and produce false positives.

[lh3]

Actually, it is not always necessary to draw a line between variant detection and association. In additional to samtools, there is another just published paper doing association directly from the sequencing data.

[Larry_Parnell]

Although we are fans of PLINK, +1 for a clear and concise drawing of the line from variant detection to the subsequent association study. Very nice!

[Larry_Parnell]

Although we are not fans of PLINK, +1 for a clear and concise drawing of the line from variant detection to the subsequent association study. Very nice!

[David Quigley]

Can you elaborate there? Aren't you still asking whether there is a distribution of alleles that suggests that two sample sets are not drawn from the same underlying population, and that the samples segregate according to some a priori condition (e.g. cases vs. controls). That's an association study.

[lh3]

Association test can be done without genotypes. To get the allele frequency, you do not need genotype in principle.

Answer 2

[Sean Davis]

First, I would not recommend bowtie for this type of task as it is not a gapped aligner. Second, have a look at the mpileup help page and specifically at the section on performing association tests.

Answer 3

[Travis]

Thanks for the input guys!

It would be a responder/non-reponder type dataset. Would use of mpileup be sufficient?