Genetic rsisk score, polygenic risk score

Question

[J.F.Jiang]

Hi all, I am dealing with couple of SNPs to predict the risk. As suggested by several references, wGRS or PRS will be the good choice. However, most refs used quantile 1 as average risk (20% proportion), quantiles 2-4 as intermediate risk, and quantile 5 as high risk. I am confusing that if the incidence of the disease is less than 1%, how can we say 20% individuals will be at high risk? Is there any other better soltion?

An other question is that some alleles were reported as risk alleles with <1 OR, which means the allele is "protective", should we inverse the OR and change the other allele as risk to calculate the GRS? If not, I think the category of outcomes should contain low risk, which means the outcomes should be grouped into low risk, below average risk, average risk, above average risk, and high risk?

Could any one give me some suggestions?

Answer 1

[bikash.k]

1. Though incidence of the disease is less than 1%, out of 100 people with risk allele 20 individual will manifest the disease.
2. Most of the time, assigning association of SNP (risk or protective) is based on OR (<1 protective, >1 risk), however based on type of algorithm used to derive OR, the inference may vary. Therefore despite of OR<1, allele might be risk. Hence, always look up to distribution (frequency) of allele among Case and Control.