Daniel Gustavson has computed polygenic scores (PGSs) using a standardized pipeline.

THIS PAGE IS STILL UNDER CONSTRUCTION. IF YOU HAVE QUESTIONS PLEASE EMAIL daniel.gustavson@colorado.edu

Plink files for computing PGSs in CATSLife can be found on Research Computing: /pl/active/IBG/data/CATSLife/derived/PGS/

We started with the lightly QC'd genetic datafiles from Luke Evans that was imputed to HRC. Additional QC was then applied to remove INFO/R2<.80, remove multiallelic variants, and rename to RSID (build 37). Additional filters in plink included: –geno .05 –maf .01 –mind .05 –hwe 1e-10

PGSs were then computed using PRScs (Ge et al., 2019) and/or SbayesR (Lloyd-Jones et al., 2019) software. See the table below for which specific traits have been scored. Example scripts for generating your own scores are provided on RC, but please contact Daniel Gustavson if you have any questions and/or want to request specific scores generated in the future.

Information about Genetic Ancestry and Principal Components can be downloaded here and can also be found on RC: /pl/active/IBG/data/CATSLife/derived/PGS/

The current file includes important pieces of information recommended for PGS analyses.

• Conversion from genetic IIDs to standard CATSLife IDs needed to merge PGS output with phenotypic information.
• Continental ancestry group. These were determined based on mapping CATSLife data to the 1000 Genomes Phase 3 reference panel. EUR (European-like), AFR (African-like), SAS (South Asian-like), and EAS (East Asian-like) groups were classified if they were within 5 SDs of the population means based on the first 4 principal components. Because there is a lot more variability in the AMR (American-like) group, this based on 2 SDs on the first 4 principal components and required that they not already be classified in one of the other groups.
• Two sets of PCs are provided in this datafile. Principal components labeled “PC1”, “PC2” (etc.) are derived from a PCA within this sample. Principal components labeled “PC1_1KG”, “PC2_1KG” (etc.) are derived from mapping our data onto 1000 Genomes. For polygenic scores involving CATSLife data, it is recommended that you use the within-sample PCs, though there may be exceptions.

Readme for PRScs (Ge et al., 2019)

Readme for SbayesR (Lloyd-Jones et al., 2019)

* (Restricted to 23andMe agreement, contact Dan Gustavson & Naomi Friedman)

Download merged PGSs computed using PRScs