A Toolkit Supporting Formal Reasoning about Causality in Metabolic Networks

Citation

Bodei C, Bracciali A & Chiarugi D (2007) A Toolkit Supporting Formal Reasoning about Causality in Metabolic Networks. Nettab 2007 Workshop: A Semantic Web for Bioinformatics: Goals, Tools, Systems, Applications, University of Pisa, Italy, 12.06.2007-15.06.2007. http://www.di.unipi.it/~braccia/DATA/MY_PAPERS/NETTAB_BBC_07.pdf

Abstract
Background: Metabolic networks present a complex interconnected structure, whose understanding is in general a not trivial task. Several formal approaches have been developed to support the investigation of such networks, like quantitative models based on ODEs and machine learning techniques. One of the relevant problems in this context is the comprehension of causality dependencies amongst the molecules involved in the metabolic process. Results: We propose a formal analysis approach aiming at featuring both expressiveness and ease of use. Its main ingredients are: i) a minimal notation to precisely represent bio-chemical interactions, and ii) an automated tool allowing the human expert to easily vary conditions of the in silico experiment. In particular, we exploit an analogy between logical implication and chemical reaction, i.e., roughly, the reaction of two molecules A and B producing a third one, C, can be interpreted as A and B logically imply C. Starting from a description of a metabolic network, in terms of reaction rules and initial conditions, chains of reactions, causally depending one from the another, can be mechanically deduced. Then, both the components of the initial state and, noticeably, the clauses ruling reactions can be changed and a new trial of the experiment started, according to a what-if investigation strategy. The method is supported by a computational logic counterpart, based on a Prolog implementation, which allows for a representation language closely correspondent to the adopted chemical abstract notation. The proposed framework has been validated by studying the robustness of the metabolic network of Escherichia coli K12. Selected genes have been knocked-out by disabling the rules regarding the encoded enzymes. Results are coherent with the actual biological behaviour. Conclusions: Starting from the presented work, our goal is to provide an effective analysis tool, supported by an efficient full-fledged computational counterpart, which can fruitfully drive in vitro experiments by effectively pruning non promising directions. More large-scale experiments are ongoing.