The process of metabolic network reconstructions typically begins with the task of collating existing information. For metabolites this poses a relatively straight forward set of cheminformatics problems. This workflow implements matching algorithm M2 which reads molecules from two sources and produces clusters of highly similar molecules.