Microhomology-mediated break-induced replication (MMBIR) is a DNA double-strand break (DSB) repair pathway that proceeds by polymerase template switching between sites of microhomology and results in templated insertions and complex genomic rearrangements. Such microhomology-flanked insertions were implicated in promoting neurological diseases and cancer. Despite their potential clinical relevance, MMBIR mutations have historically been understudied as their intermediate size (~10-50 bp) makes them challenging to detect by standard next-generation whole genome sequencing (WGS) pipelines. To identify MMBIR insertions we developed a novel bioinformatic tool – MMBSearch. Using this tool, we showed that healthy individuals do not accumulate de novoMMBIR mutations in their fibroblasts or blood. Meanwhile, MMBIR mutations accumulate substantially in various cancers including ovarian, colon, cervical, and lung cancer.

Our recent results using MMBSearch to analyze whole exome sequencing (WXS) datasets showed improved MMBIR identification due to less noise. Using WXS, we identified hundreds of de novo MMBIR mutations in melanocytes in contrast to the handful in fibroblasts from the same healthy individuals. WGS analysis of cancers – including melanoma, glioblastoma, and uterine – showed abundant tumor MMBIR mutations, often exceeding 1,000/exome. Interestingly, some patients also had many MMBIR mutations in their matched control samples indicative of individual predispositions.