It has been shown that combining rare and common genetics variants improves population risk stratification for breast cancer. Several models have been defined to combine PRS with rare truncating variants on a short list of genes for which robust breast cancer risk estimates are available. However, there is evidence that missense variants confer elevated breast cancer risks and that there are other genes for which the association with breast cancer was established (eg. TP53, PTEN, STK11, CDH1, NF1 and NBN). Therefore, using a Bayesian approach, we aimed to select rare LoF and Missense rare variants and combine them to define a score that, added to PRS, improves the ability to predict the occurrence of breast cancer.