Reasoning over biological networks using maximum satisfiability

  • Authors:
  • João Guerra;Inês Lynce

  • Affiliations:
  • INESC-ID/Instituto Superior Técnico, Technical University of Lisbon, Portugal;INESC-ID/Instituto Superior Técnico, Technical University of Lisbon, Portugal

  • Venue:
  • CP'12 Proceedings of the 18th international conference on Principles and Practice of Constraint Programming
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Systems biology is with no doubt one of the most compelling fields for a computer scientist. Modelling such systems is per se a major challenge, but the ultimate goal is to reason over those systems. We focus on modelling and reasoning over biological networks using Maximum Satisfiability (MaxSAT). Biological networks are represented by an influence graph whose vertices represent the genes of the network and edges represent interactions between genes. Given an influence graph and an experimental profile, the first step is to check for mutual consistency. In case of inconsistency, a repair is suggested. In addition, what is common to all solutions/optimal repairs is also provided. This information, named prediction, is of particular interest from a user's point of view. Answer Set Programming (ASP) has been successfully applied to biological networks in the recent past. In this work, we give empirical evidence that MaxSAT is by far more adequate for solving this problem. Moreover, we show how concepts well studied in the fields of MaxSAT and CP, such as backbones and unsatisfiable subformulas, can be naturally related to this practical problem.