Computational tools for modeling heterogeneity within cell populations
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Sometimes cells within the same population exhibit different characteristics, even when they have the same DNA. This non-genetic heterogeneity is not only interesting as a fundamental problem in biology, but also relevant from a human health standpoint. For example, some cancer cell populations switch back and forth between transient phenotypic states, and cells in one state survive drug exposure, while cells in the other state do not. We introduce a computational method for estimating the rates of state-switching in cell populations based on data from a Luria-Delbrück style fluctuation test experiment. We then benchmark this method on data from computer simulations of the experiment, and show that it is reasonably effective and robust to various assumptions about the cell cycle time distribution. One cause of non-genetic heterogeneity in cell populations is stochastic variation in gene expression, combined with complex networks in which genes regulate the expression of other genes. A widely studied problem in computational biology is the attempt to infer the structure of gene regulatory networks from transcriptomic datasets, called "network inference." This problem has been studied for more than a decade and dozens of network inference methods have been developed, but it is still unclear how well they work when applied to real datasets. Attempts to benchmark these methods on experimental datasets have yielded mixed results, in which sometimes even the best methods perform no better than random guessing, and in other cases they perform reasonably well. We investigate through simulations and mathematical analysis how network inference from transcriptomic data may be more or less feasible under different noise conditions in gene regulatory relationships. We also benchmark network inference methods on experimental and simulated datasets to evaluate their level of accuracy. Finally, we apply network inference techniques to experimental data from melanoma cells in an attempt to study the phenomenon of epigenetic reprogramming and drug resistance.
Description
Keywords
Melanoma cells, Genetic heterogeneity, Computational biology, Network inference, Drug resistance