our previous studies14–16. Briefly, Individuals and SNPs with genotype call rates <98%, and SNPs with minor allele frequency <1% and Hardy–Weinberg equilibrium P < 1 × 10−6 were removed from downstream analyses. After the pre-imputation quality control, genotype data were imputed using Minimac317 implemented in the Michigan Imputation Server (available at https://imputationserver.sph.umich.edu/) with the 1000 Genomes Phase 3 reference panel18. Dosage data were transformed into best-estimate genotypes using PLINK219, considering variants with info score ≥80% and minor allele frequency ≥1%. In the Yale-Penn participants of African and European descent, we investigated 4,915,647 and 4,202,333 variants, respectively. The present study only considered information regarding unrelated subjects. As previously described20, individuals with an identity-by-descent proportion >0.125 were defined as belonging to the same family group. Within each family group determined from genetic data, a subject was selected prioritizing retention of participants who reported the most extreme suicidality among those genetically related.
