simple polygenic risk example
0
0
Entering edit mode
5.6 years ago
nkinney06 ▴ 90

I am trying to calculate polygenic risk scores. Basically these are calculated using a set of weighted SNPs. The score assigned to an individual can be correlated with many things such as risk for disease. Many polygenic risk scores (or models) have been devised and published in literature. I though I would be able to look up one of these publications and find a list of the SNPs they use along with the weights and formula for calculating the score. That way I could use their model to calculate polygenic risk for an unknown sample. However, despite the many publications, I am finding it hard to find any that present their model so simplistically. For some papers the calculation of polygenic risk score is hopelessly impenetrable. Are there any review articles that make calculating polygenic risk more approachable? any there any papers that propose a polygenic risk model (for anything) in simple terms with the SNPs, weights, and formula for actually applying the model? How do I look up a published model and calculate the score for an unknown sample? Why dont these publications come with a script for computing the score given a bam file or vcf file? Any help is appreciated.

SNP polygenic risk • 3.4k views
0
Entering edit mode

Just checking to see if there might be some responses to these question, as it's now three years later, and I'm finding the same thing...

2
Entering edit mode

My understanding is that usually you will have a list of SNP genotypes with each having an odds ratio determined by comparing a control group and experimental group of samples (usually 1000s of samples). You take the log of all the odds ratios and sum the logs; so, for an unknown sample check each SNP in the test of polygenic risk and sum the log odds of each genotype. Usually there will be some cutoff determined via ROC analysis. If the unknown sample is above the cutoff it is predicted to match the experimental group. Supposedly the sum of log odds is a fundamental result in genetics but I have not seen the derivation (I would like to); probably look into regression analysis. I had an idea awhile back about comparing polygenic risk scores. I gave up because I found too many papers with home-cooked methods and/or unclear methods. I ended up going in a different direction so I could even be wrong about some of what I say. If you find answers and/or clear derivation let me know!