We developed two set of scripts to help us consistently deal with two kinds of problems that we encountered in building the database:
In this category, we simply find that there are two separate compounds that actually represent the same identical molecular structure and, for whatever reason, were not merged when building the database. Here we merge them. Doing so involves merging all the attached data for both of them, and rendering one of them obsolete for the purposes of metabolic modeling.
In this category, we find that two compounds from the external sources that were used to build the database were incorrectly merged, for whatever reason (a common reason is a generic synonym). To curate this problem, we wrote scripts that separated out the two original compounds into two separate entities, and sorted out the associated data, espcially aliases and structures. Sometimes, this means creating an entirely new compound, and sometimes, the second compound already exists, in which case we'd merge the associated data with it. The process of separating the data out, using provenance where we can is convoluted and we describe it a little more in the Disambiguation directory.