How to identify which SNPs in my dataset are genome wide significant

Question

[sean.oscineir2]

Hi

Unfortunately I've missed out on one or two of my practical's for GWAS and the screencasts don't touch on this.

I currently have two SNPs that are potentially significant but I haven't a breeze on how to identify what the SNPs actually are based off the graphs (A manhattan plot and a QQ-plot)

What am I missing in regards to obtaining SNP ID?

Apologies for the silly question

https://imgur.com/vA9c66g https://imgur.com/f3MkcVW

[WouterDeCoster]

How to add images to a Biostars post

[h.mon]

Can a SNP be genome-wide significant? What exactly do you mean?

[sean.oscineir2]

Not single SNPs however the presence of multiple SNPs associated with disease can be genome wide significant, at least that's my take home from it

This assignment has a small dataset so the two shown in the Manhattan plot have some degree or significance

All I'm trying to figure out is how to get the SNP ID for the two on the Manhattan plot but I'm not sure where I'm finding it in my dataset. The only information have significant values for is heterozygosity and missingness rate

Answer 1

[Johan Zicola]

It seems your QQplot and Manhattan plot were generated with the R package qqman. You can filter out the SNPs above the threshold you define, the blue and red lines being set by default at -log10(1E-5) and -log10(5E-8) (check documentation). However, it could make more sense to use a threshold with a Bonferroni correction (alpha-risk / number of SNPs tested).

In your case, let's say you want to extract all SNPs above the red line for your 2 SNPs:

# GWAS results are in the dataframe gwas.results which contains 4 variables: SNP, CHR, BP, P

# Get SNPs having a p-value inferior to 5E-8
threshold <-  5E-8
gwas_significant <- subset(gwas.results, P < threshold)

# Display SNP IDs
gwas_significant$SNP