Matching Causality Hypotheses to Simulations of Biological Systems

Abstract

Computational models for studying chemical reaction networks are indispensable tools for analyzing the intertwined reactions involved in complex biological systems. Indeed, executable models such as the programming language Kappa provide useful insights by allowing users to specify a system of chemical rules, and then simulating plausible sequences of reactions that might execute in our cells given this rule system. Kappa thus provides the opportunity to analyze the sequence of steps that lead to a chemical event of interest. Formal models for the causal structure of these sequences have been developed based on Kappa, describing the chain of events that directly or indirectly contribute to the production of an event of interest. These models, known as stories, attempt to capture the notion of a biological pathway. In this thesis, we develop a novel framework for matching stories to Kappa simulations, ensuring that we report a match if and only if the story is a reasonable model for the underlying causality of the simulation. We then present algorithms to efficiently compute these matches, and apply these algorithms in two settings: we study the prevalence of competing pathways in cellular death, and we analyze the dynamics of a common biological pattern, three-membered ring formation. Our insights on these systems align with those presented in literature, demonstrating that these matching algorithms have applicability for analyzing biological systems.