I am calculating PRS scores from the PGS Catalog with their tool "pgsc_calc", and I am a bit confused about how to interpret them.
The specific PGS I am using is PGS002765, which has "beta" as effect weight.
When I calculate the score for an individual and get a score of: 1.04067160633191e-7 as a result, how can I evaluate this score and say that it is either good or bad? And what does the effect weight "beta" mean?
I am no expert on this topic and thankfull for any replies. :)
•
written
by
Patrick •
PRS is a population measure. Most of the time, what you got is the score, which does not have much meaning if you only have one sample. What you need is something akin to absolute risk, which is rarely available and afaik, there are still a lot of technical challenge in getting absolute risk from PRS.
Hello,
Based on the previous discussion, I would like to inquire about the following: If I have a tested sample along with 49 general samples from the "1000 Genomes" project, making a total of 50 samples with calculated scores, what should I do next? Can I use percentile ranks to describe the risk level, or do I need to use a normal distribution or linear regression to determine the risk level?
Thank you.
Hi Leo, I made a comment below with a pointer to the methods and a tool we developed for this purpose below (Interpretation of PRS scores calculated from PGS Catalog)
You can basically convert a PGS to a relative risk score by comparing the PGS to a reference distribution. There are multiple ways to do this (see https://pgsc-calc.readthedocs.io/en/latest/explanation/geneticancestry.html) by comparing to the scores of genetically similar people or by using PCA to regress out the effects of population structure. You may find our PGS Catalog Calculator tool useful for this (https://github.com/pgscatalog/pgsc_calc), and we provide a copy of HGDP+1kGP as a reference population panel.
Log in to answer this question.
