DESCRIPTION (provided by applicant): Genome-wide association studies have been very successful in identifying hundreds of variants associated to complex diseases and phenotypes. In contrast, due to high levels of linkage disequilibrium at any given locus, only a handful of causal variants have been identified so far. In an attempt to bridge this gap, several fine- mapping studies involving dense genotyping or sequencing are currently being performed in multiple populations such as Europeans, Asians, African Americans or Latinos. Fine mapping studies over multiple populations can leverage different genetic variation across populations to increase the accuracy for localizing the causal variant in a joint analysis of multiple populations as compared to studies in which only one population is analyzed at a time. Surprisingly, despite the large potential of multi ethnic fine mapping studies, current multi population fine mapping studies employ standard statistical techniques within locus specific ad- hoc frameworks. In this application we will introduce novel metrics and automated frameworks for quantifying the performance of fine mapping methods as well as novel statistical methods that leverage multi ethnic genetic variation to increase the localization accuracy for fine mapping.

Resistance to a wide range of cancers, including breast cancer and various other diseases, is known to include a substantial genetically heritable component. Genome wide association studies have been very successful in identifying loci associated to various diseases including breast cancer. In contrast, the underlying genetic causal variants have yet to be identified for large number of phenotypes including most cancers. In this application, we will develop novel methods and metrics for multi-ethnic fine mapping studies and apply them to real fine mapping breast cancer data sets.