al.61173,005 (59,851/113,154)4,924,025850,29115,276 (2361/12,915)Inflammatory bowel diseaseIBDLiu et al.6234,652 (12,882/21,770)4,823,570849,74918,998 (750/18,248)Rheumatoid arthritisRAOkada et al.6358,284 (14,361/43,923)3,872,637849,68018,170 (753/17,417)Type 2 diabetes mellitusT2DMScott et al.64159,208 (26,676/132,532)4,901,848856,91218,823 (1978/16,845)HeightHGTYengo et al.65693,5291,578,533750,8883957Body mass indexBMIYengo et al.65681,2751,579,905751,6763954High-density lipoproteinsHDLWiller et al.66188,5781,604,577758,0362491Low-density lipoproteinsLDLWiller et al.66188,5781,600,625756,7241713CholesterolCHOLWiller et al.66188,5781,604,391757,9702561TriglyceridesTRIGWiller et al.66188,5781,601,270756,9132505The sample size for each external genome-wide association study (GWAS), and the number of genetic markers included in the polygenic prediction are shown, along with the sample size for each disease and quantitative phenotype in the Partners HealthCare Biobank (PBK). For unadjusted PRS and P+T, all common genetic markers (minor allele frequency ≥1%) that passed quality control and are present in the summary statistics and 1000 Genomes Project (1KG) European sample were used in prediction. For LDpred(-inf) and PRS-CS(-auto), genetic markers were further restricted to the HapMap3 (HM3) panel
