(430e) MC3: A Tool for Model and Constraint Consistency Checking of Stoichiometric Biochemical Network Models | AIChE

(430e) MC3: A Tool for Model and Constraint Consistency Checking of Stoichiometric Biochemical Network Models


Yousofshahi, M., Tufts University
Stern, R., Tufts University

Stoichiometric models are traditionally constructed manually, based on earlier models, the availability of reaction data from databases, and newly available experimental data. An example is the formulation of successive models of the Escherichia coli organism that lead to improved predictive capabilities and elucidation of phenotypic behavior. Models thus have gaps (missing reactions), and there is sometimes uncertainty about reaction directions.  While some methods have been advocated to ensure the correctness of some aspects of a stoichiometric model within a particular context, there is currently no standalone computational tool that ensures model and constraint consistency.  Each user is thus forced to manually sanitize his or her model, and often it is an incorrect computational result that alerts the user of the incorrectness of the model.  Models released in the public domain often have undocumented issues, such as dead-end metabolites or zero-flux reactions.  Additionally, there is no standard documentation protocol that each model undergoes prior to public release.  The need for agreed-upon standards and for computational tools will become more pressing with the rise of the size and number of genome-scale models.

We define and formalize the notion of a valid stoichiometric biochemical network model suited for steady-state analysis, and develop a computational tool, MC3 tool, a Model and Constraint Consistency Checker, to validate stoichiometric models. The MC3 tool consists of two distinct checking components, one based on the results of the kernel computation for S .v =0 and one based on exercising Flux Balance Analysis.  We checked the consistency of some published available models using MC3. The checked models included those of Escherichia coli (3 different sizes), adipocyte, Chinese Hamster Ovary (CHO) cell, Leishmania major, and Zymomonas mobilis.  For every test case, MC3 checks for the dead-end metabolites, zero flux reactions, unsatisfied reversibility and inconsistent irreversibility. Our tool showed several issues with current released models including inconsistent documentation of dead-end metabolites, models with a considerably large number of undocumented zero-reaction fluxes and a number of inconsistently reversible reaction pairs.  The MATLAB tool has been placed in the public domain on google code (http://code.google.com/p/mc3/) in April 2012.  We hope that this presentation will raise awareness about the need to standardize the documentation of genome-scale stoichiometric models.

See more of this Session: In Silico Systems Biology: Cellular and Organismal Models II

See more of this Group/Topical: Topical A: Systems Biology