Additional file 1 of Genome-wide prediction of pathogenic gain- and loss-of-function variants from ensemble learning of a diverse feature set

Additional file 1: Table S1. Number of variants per class in complete, training, and testing datasets. Table S2. Description of features used for LoGoFunc’s development. Table S3. Fisher's exact test results comparing the GOF, LOF, and neutral variants across protein features. Table S4. Odds ratios and log odds ratios comparing the GOF, LOF, and neutral variants across protein features. Table S5. Fisher's exact test results comparing the GOF and LOF variants across protein features. Table S6. Odds ratios and log odds ratios comparing the GOF and LOF variants across protein features. Table S7. GOF, LOF, and neutral prediction 95% confidence intervals per gene. Table S8. Feature importance values for features used to train LoGoFunc. Table S9. PheWAS using predicted GOF and LOF missense variants in the BioMe BioBank cohort. Table S10. Replicated PheWas results on an independent BioMe WES cohort.