Fig. 1: Generalizability of bidirectional effects for increasing probability of success on clinical trials.

A Graphical representation of a gene (PCSK9) showing bidirectional effects on LDL and cardiovascular disease risk. B Circos plot of 98 genes with bidirectional phenotypic effects. From outside to inside: all autosomes and the X chromosome are depicted in a clockwise orientation. The genomic coordinates are shown for each gene that was identified to contain different variants with phenotypically opposing HGMD annotations. The panels with radial lines display the total number of HGMD variant entries (log base 2) that were categorized as bidirectional, for each gene. Each gene label is color coded based on the phenotype that displayed bidirectionality. C Estimated odds ratio for transitions between various clinical trial phases. We calculated the odds ratio for transition and 95% confidence interval, for the subset of target-indications with genetic support from GWAS, OMIM associations with unidirectional effects, and bidirectional effects curated from HGMD. Measure of center for the error bars represent the estimated odds ratios and the error bars represent the 95% confidence intervals. Bidirectional effect supported data is shown in gray, OMIM unidirectional support is shown in orange, and GWAS support is shown in blue. n = 26,884 target-indication pairs.