In addition, we performed the multivariate PRS and the Cox’s proportional hazard analyses of recurrent depression using the ancestry outlier sample (Supplementary Figure S25-S28 and Supplementary Table S25-S28D), adjusting all analyses for the first 10 principal components and iPSYCH2012/iPSYCH2015i. The ancestry outlier mvPRS analyses (Supplementary Figure S25) showed similar directions of differences in DEP-PRS and Neuroticism-PRS for single vs. recurrent depression as in the corresponding analysis of the sample with European ancestry (see Supplementary Figure S16-1, Supplementary Table S16A), although none of the differences appeared to be significant (Supplementary Table S25). For ANX-PRS and BP-PRS, however, the observed differences in the European sample were not reproduced for the ancestry outliers. Likewise, for the Cox’s regression analyses, the observed differences in HRR and absolute risk of developing a second episode of depression between10th and middle DEP-PRS and DEP-SUM-PRS deciles in the European ancestry sample (Supplementary Figure S17-1) were not reproduced in the ancestry outliers (Supplementary Figure S26). Regarding psychiatric comorbidity among depression cases a similar pattern of increased load of ANX-PRS, SZ-PRS and Neuroticism-PRS among depression cases with comorbidity for anxiety
