Comparison of SNPs between groups of samples

Question

[agata88]

Hi all!

Could you please recommend a tool that can statistically compare SNPs between two groups of vcf's?

To be more clear, I have 30 vcf files for healthy patients, and 30 for sick patients. I need to search for some kind of SNP patterns, statistically significant between those two groups.

Any idea how to perform such analysis?

Thanks in advance,

Agata

Answer 1

[Pierre Lindenbaum]

run a fisher test for those variants. https://en.wikipedia.org/wiki/Fisher%27s_exact_test

[agata88]

Thanks, it's one way. I'am thinking about something different.

What is on my mind is to go back to variant detection step and use VarScan Somatic caller. I might use samtools mpileup to generate two vcf's each containing variants for 30 samples. Do you think it's good idea?

[Pierre Lindenbaum]

what would be the

"kind of SNP patterns,"

here ?

[agata88]

I would like to see whether there are variants that occur only in sick patients and not in healthy group. In addition, I would like to see that in multiple samples comparison. For example, let's say I've found 5 mutations that occur in 20 sick samples and not occur in healthy group at all - this 5 mutation will create some kind of pattern.

[Pierre Lindenbaum]

and you'll need a statistical test ...

[agata88]

to show significant difference at p< 0,05.