David McCandlish

The McCandlish lab develops computational and mathematical tools to analyze and exploit data from high-throughput functional assays. The current focus of the lab is on analyzing data from so-called “deep mutational scanning” experiments. These experiments simultaneously determine, for a single protein, the functional effects of thousands of mutations. By aggregating information across the proteins assayed using this technique, we seek to develop data-driven insights into basic protein biology, improved models of molecular evolution, and more accurate methods for predicting the functional effects of mutations in human genome sequences.

Critically, these data also show that the functional effects of mutations often depend on which other mutations are present in the sequence. We are developing new techniques in statistics and machine learning to infer and interpret the complex patterns of genetic interaction observed in these experiments. Our ultimate goal is to be able to model these sequence-function relationships with sufficient accuracy to guide the construction of a new generation of designed enzymes and drugs, and to be able to predict the evolution of drug resistance phenotypes in both populations of cancer cells and rapidly evolving microbial pathogens.