Model Revision of Boolean Logical Models of Biological Regulatory Networks

Abstract

Complex cellular processes can be represented by biological regulatory networks. Computational models of such networks are essential to have a better understanding of the corresponding cellular processes, allowing the reproduction of known behaviours, the testing of hypotheses, and the identification of predictions in silico. However, the construction of these models is still mainly a manual task, and therefore prone to error. Additionally, as new data is acquired, the existing models must be revised. Here, we propose a model revision procedure, capable of providing the set of minimal repairs to render a Boolean logical model of a biological regulatory network consistent with a set of experimental observations. The proposed model revision procedure takes advantage of a logic-based approach to verify whether a model is consistent with a given set of experimental observations. In case of inconsistency, the consistency check procedure implemented using Answer Set Programming (ASP) determines the minimum number of inconsistent nodes of the model and corresponding reason of inconsistency. An algorithm to search for possible sets of repair operations to render an inconsistent model consistent is proposed. In this work, four repair operations are considered: changing a regulatory function; changing a type of interaction; removing a regulator; and adding a regulator. The model revision approach presented here, considers confronting a Boolean logical model with stable state or time-series observations. Moreover, for time-series observations, both synchronous and asynchronous update schemes are considered. The proposed model revision approach is tested on five published well known biological models. Corrupted versions of these models are generated by performing random changes. The corrupted models are confronted with stable state observations, corresponding to the stable states of the original models, using the presented model revision procedure. Moreover, different time-series observations are generated, consistent with the original models, to assess whether the model revision approach is able to revise the corrupted model under time-series observations. The proposed method is able to repair the majority of the corrupted models, considering stable state and time-series observations. Moreover, all the optimal solutions to repair the inconsistent models are produced.

Filipe Gouveia

Computer Science Researcher

My research interests include artificial intelligence, computational logic and automated reasoning.