The results of the present study demonstrate how semi-automatic gap filling can be used to refine and extend metabolic reconstructions, thereby increasing their biological scope. We subsequently experimentally validated one of the proposed hypotheses and report that C9orf103, previously identified as a candidate tumour suppressor gene, encodes a functional human gluconokinase. Through detailed case studies we show that biologically plausible non-intuitive hypotheses regarding the metabolism of these compounds can be proposed in a semi-automated manner, in an approach that is similar to de novo network reconstruction. MetaCyc is also used as a knowledge base. Solutions were characterized with respect to how metabolites were incorporated into RECON 1 and their biological relevance. The data in MetaCyc are evidence-based and richly curated, resulting in an encyclopedic reference tool for metabolism. Candidate solutions, composed of metabolic reactions capable of explaining the metabolism of these compounds, were then identified computationally from a global biochemical reaction database. In the present study we used a computational approach to identify human metabolites whose metabolism is incomplete on the basis of their detection in humans but exclusion from the human metabolic network reconstruction RECON 1. Metabolic network reconstructions define metabolic information within a target organism and can therefore be used to address incomplete metabolic information.

Universal database metabolic reactions