Multiple Biological Model Classification: From System Biology to Synthetic Biology

Abstract

This paper describes how to classify a family of biochemical pathways and circuits in terms of their temporal behavior by systematic application of time-frequency analysis and temporal logic based model checking. There are two immediate pay-offs for this approach. First, a family of models obtained by systematic perturbations of an archetypal (e.g., wild-type or ancestral) model, when classified in this manner, can help in identifying various incorrect or implausible features of the model. Second, specific hypotheses about various features of a given model can be automatically generated from such analysis and then subjected to experimental verification. We illustrate, by two examples, how our approach can be used (1) to understand the behavior of any individual topologically distinct circuit among the set of 125 synthetic biological circuits created out of similar elements, or (2) to ascertain the correctness of a well known Yeast Cell Cycle model by checking a family of perturbed models created through single and double mutations. We also show how this tool interacts with a propositional temporal logic model-checking system to present qualitative distinctions among the groups within the family of biological circuits or among the different multi-modal behaviors of a single pathway. The paper also poses several challenging open problems in computer science, related to our basic generate and test framework (with heuristics) used for systematically producing a set of temporal logic sentences that can be tested against different datasets.