Specific combinations of genes exhibit emergent properties when expressed together, enabling the generation of diverse cell types and behaviors. This phenomenon motivates the quantitative study of genetic interactions (GIs), which compare the phenotypic consequences of perturbing two or more genes alone or in combination. The challenge of studying GIs is their sheer scale: e.g., among 10,000 genes there are ~50 million possible pairwise GIs. The Norman Lab blends computational approaches with high-throughput experimental methods to develop new screening approaches for finding and characterizing genetic interactions.

We have found particular success through the development of the Perturb-seq approach, which enables any CRISPR-mediated genetic perturbation to be identified during single-cell RNA sequencing, making it possible to connect thousands of genetic perturbations to their transcriptional consequences in pooled format. These rich phenotypes provide both deep mechanistic insights into the nature of GIs and also a “handle” to apply modern machine learning approaches. Current research focuses on experimental studies of the global structure of GIs in tumorigenesis, the development of new computational methods for designing and interpreting very large single-cell experiments, and technological approaches for studying cell-cell interactions.