Fig. 1

Odds ratios and alpha estimates for each of five genes in population training samples. A ATM. Odds ratios for breast cancer risk from logistic regression models. Alpha is the estimated proportion of risk associated variants from mixture models, based on variants in control samples. ATM risk categories: variants lying within the FAT or PI3K/PI4K protein domains with CADD score in the fifth quintile (FAT/PIK + CADD5); variants lying within the FAT or PI3K/PI4K protein domains with CADD score in any of the first four quintiles (FAT/PIK + CADD1-4); variants lying outside the FAT and PI3K/PI4K protein domains (Outside FAT/PIK). B BRCA1. Odds ratios for breast cancer risk from logistic regression models. Alpha is the estimated proportion of risk associated variants from mixture models, based on variants in control samples. BRCA1 risk categories: variants lying within the RING or BRCT domains with a high Helix score (RING/BRCT + Helix-high); variants lying with the RING or BRCT domains with a low Helix score (RING/BRCT + Helix-low); variants lying outside the RING and BRCT domains (Outside RING/BRCT). C BRCA2. Odds ratios for breast cancer risk from logistic regression models. Alpha is the estimated proportion of risk associated variants from mixture models, based on variants in control samples. BRCA2 risk categories: variants with a high Helix score (Helix-high); variants with a low Helix score (Helix-low). D CHEK2. Odds ratios for breast cancer risk from logistic regression models. Alpha is the estimated proportion of risk associated variants from mixture models, based on variants in control samples. CHEK2 risk categories: variants with a high Helix score (Helix-high); variants with a low Helix score (Helix-low). E PALB2. Odds ratios for breast cancer risk from logistic regression models. Alpha is the estimated proportion of risk associated variants from mixture models, based on variants in control samples. PALB2 risk categories: carriers of any missense variant (Carriers)