Methods for bounding genetic nonlinearities

Abstract

Complex hierarchical structures are a hallmark of life. Within multicellular organisms, the building blocks of these structures are cells; within cells, they are genes. The interdependence of these building blocks is difficult to measure but is integral to the biological processes of health and disease, which emerge from the dynamism of thousands of interacting genes. This cooperativity manifests in particular mutations which accumulate over the course of cancer progression, gender-specific medical conditions, and transcription factor cocktails used to reprogram differentiated cells into stem cells. However, it is experimentally intractable to test the significance of perturbing every unique combination of genes. Instead, we explore gross features of this interaction space to determine how prevalent these synergies are. We take a top-down approach, creating new methods to measure the effects of removing genes from the full set. In the first, we develop a method to measure the transcriptional response to genetic perturbations across hundreds of thousands of cells revealing opposing classes of transcription factors regulating the immune response of dendritic cells. In the second, we create a method to measure how millions of combinations of genetic perturbations impact the growth rate of cancer cell lines.