Women with germline inactivating mutations in the BRCA1 and BRCA2 genes are at significantly elevated risk of breast and ovarian cancer. Clinical genetic testing for mutations in these genes has become an important part of clinical practice because of the surgical prevention and targeted treatment benefits associated with knowledge of the presence of a cancer predisposing mutation. However, this process is often complicated by the detection of Variants of Uncertain Significance (VUS), which are predominantly missense mutations with an unknown influence on protein function and unknown association with cancer risk. The lack of information about these VUS means that individuals found to carry these variants cannot benefit from enhanced risk assessment or make informed decisions about surgical prevention or tailored treatment options such as platinum and PARP inhibitor therapy. Here we propose to determine the cancer relevance of VUS found throughout the BRCA1 and BRCA2 genes by developing a comprehensive model incorporating new genetic and functional approaches for VUS classification. However, because BRCA1 and BRCA2 are large proteins, with several established distinct functions that may or may not have a role in cancer development, the contribution of the different functions to tumor suppression and cancer risk must be determined before assays can be applied to VUS classification. Thus, in Aim1 we will perform a comprehensive functional analysis of BRCA1 variants to determine which molecular functions contribute to tumor suppression and in Aim 2 we will perform a comprehensive functional analysis of BRCA2 variants to determine which molecular functions influence the risk of cancer. Specifically, we will evaluate the influence of known pathogenic and nonpathogenic variants on defined functions of BRCA1 and BRCA2 and then extend our analyses to VUS using the assays found to be associated with cancer risk. In Aim 3 we propose to extend the multifactorial model for classification of BRCA1 and BRCA2 VUS. Here we will develop new methods allowing incorporation of personal and family histories of VUS probands into an established predictive model. We will apply the model to classification of candidate moderate risk VUS, and most importantly, we will develop a Bayesian mixture model that integrates quantitative functional assay data with genetic data for the purposes of VUS classification.