UC San Diego

A grand challenge in biology is to unravel the complex relationship between genotype and phenotype. Here, I describe a systematic genotype-to-phenotype mapping platform based on combinatorial CRISPR/Cas9 to identify genetic interactions in cancer cells and a biologically inspired, deep learning method to predict and generalize these data types.

First, we interrogate essential functions and their context dependencies using ~6 million combinatorial gene disruptions in breast, lung, and oropharyngeal tumor cells. Approximately 1,800 synthetic-essential gene combinations, of which 34% are penetrant across tumor types, converge on 49 multi-gene systems. Most essential systems are identified by interactions with outside functions.

Second, we use combinatorial CRISPR/Cas9 perturbations to uncover an extensive network of functional interdependencies among CDKs and related factors, identifying 43 synthetic-lethal and 12 synergistic interactions. We dissect CDK perturbations using single-cell RNAseq, for which we develop a novel computational framework to precisely quantify cell-cycle effects and diverse cell states orchestrated by specific CDKs.

Finally, I present a visible neural network model called DCell that couples a neural network to a hierarchical structure of a cell. Trained on several million genotypes, DCell simulates cellular growth nearly as accurately as laboratory observations. During simulation, genotypes induce patterns of subsystem activities, enabling in silico investigations of the molecular mechanisms underlying genotype-phenotype associations. These mechanisms can be validated, and many are unexpected; some are governed by Boolean logic.

Together, these works describe a framework to systematically interrogate the complexity and diversity of biological functions.