FASTERCC: Accelerating Flux Consistency Testing and Context-Specific Reconstruction for Large-Scale Metabolic Network Models

Abstract

The increase in size of metabolic network models especially with the advent of single-cell data calls for scalable reconstruction and analysis tools. Such models, often used for drug discovery and the analysis of microbial communities rely on consistency testing and reconstruction algorithms such as FASTCORE and FASTCC. However, with models nowadays comprising hundreds of thousands of reactions, the running times of such algorithms increased from few minutes to hours or days even with high performance computing. Experiments that require multiple reconstructions, such as parameter tuning or cross-validation, are practically infeasible in very large networks. Here we introduce FASTERCC, a new version of FASTCC, that leverages structural information for removing type I and II dead-ends, the orientation of reversible reactions and correcting the reversibility of reactions that are structurally incapable of carrying flux in both directions prior to any feasibility tests. These improvements reduce drastically the running time of FASTERCC by a median 20-fold speedup in comparison to FASTCC for networks with a larger number of block reactions. The model cleaning performed by FASTERCC also reduces the computational time of downstream analyses, notably of FASTCORE up to 50%.