CPM Seminar

Quantifying Waddington's Landscape: Modeling the gene networks that underlie cellular identity

Pankaj Mehta

Boston University

The molecular basis of cellular identity and differentiation remains one most the fundamental questions in biology. The work of Takahashi and Yamanaka demonstrating that the overexpression of just four transcription factors is sufficient to convert fibroblasts to iPSCs has revolutionized stem cell biology. Great progress has been made in identifying the molecules and genes that shape cellular identity using modern genomic techniques such as whole-genome expression studies and epigenetic profiling. Yet, how these molecular components interact to give rise to the ‘epigenetic landscape’ underlying cellular identity remains elusive due to the large number of genes and molecules involved, the plethora of interaction between genes, and the importance of stochasticity in cellular reprogramming. Here we present a new class of probabilistic, mathematical models for the biomolecular networks underlying cellular identity using existing genomic and epigenetic data sets. The models build upon insights from the statistical physics of spin glasses and attractor neural networks and are intrinsically stochastic and large scale. Our models can reproduce many experimental observations including stable cellular states and rare probabilistic switching between cell types. The models yield new biological insights into the existence of spurious cell types (such as partially reprogrammed cells), and the importance of weak interactions for maintaining cellular identity. The ultimate goal of this research is to use these mathematical models to rationally design cellular reprogramming.