Our lab is developing DrugMechDB, a database of drug mechanisms. We are looking for one or more curators to help expand this resource with additional records. This would be a paid, remote, short-term freelance/contract role, and successful and significant contributions would result in co-authorship on a future publication. Please email Andrew (subject: “DrugMechDB curation”) if you’re interested. More details are below.
DrugMechDB is a database of drug mechanisms that are expressed as a path through a knowledge graph. Current catalogs of drug mechanisms provide annotations expressed as either drug-to-target relationships (e.g. BCR-ABL tyrosine kinase inhibitor) or as classes of actions (e.g. anticonvulsants). We aim to create a more complete picture of how a drug acts on disease by representing mechanisms as a path that begins with the drug, and includes the target, the target’s interactions with other chemicals, proteins, genes, pathways or processes, and how those ultimately relate to the disease. The goal of this catalog is to produce a set of known true mechanistic paths that can be used in the context of graph-based machine learning methods. This set of paths has the potential to inform models to predict drug-treats-disease relationships as well as identify drug-target and drug-pathway interactions important to this treatment relationship.
We’re looking for someone with expertise in formal biocuration, pharmacology and biomedical ontologies to further expand DrugMechDB. The process entails curating drug mechanisms from public data sources and manuscripts, identifying the important concepts, and encoding them as machine readable paths. Each mechanism in DrugMechDB is represented as an ordered set of interactions between biomedical concepts, annotated with stable external identifiers, that as a whole represent a path in a biomedical knowledge graph. The process of curating paths for DrugMechDB will include mapping concepts to biomedical ontologies and distilling relationships to a minimal set capable of encompassing the interaction, with primary focus on the directionality of the relationships (increases vs decreases, activates vs inhibits).