To derive a PRS for depression within the iPSYCH2015 sample we used the depression meta-analysis12,14,23,24 excluding all iPSYCH samples as training. We also calculated PRS based on GWAS summary statistics from bipolar disorder118, schizophrenia121, ADHD13 and autism15. The PRS for ADHD was based on a combination of “external” GWAS sumstats (without iPSYCH data) and “internal” training in the iPSYCH sample. The iPSYCH sample was split into 10 groups, and in 10 iterations of leave-one-out, the PRS was calculated for individuals in each of the 10 groups using a GWAS performed on the remaining 9 groups as training. A fixed-effect variance-weighted meta-analysis, implemented in METAL28, of two GWAS for anxiety, one based on self-report of physician diagnosis of anxiety in the MVP112 and the other being core anxiety in the FinnGen cohort22 (see above), served as weights for generating an ANX-PRS summarizing genetic risk of anxiety. To derive a PRS for neuroticism we used GWAS summary stats of a weighted neuroticism sum-score, constructed by adding up ten individual item responses (Fed-up, Guilt, Irr, Miss, Mood, Tense, Nerv feel, Suf Nerv, worry
