Analyzing text to extract information about gene-drug interactions

Abstract

Pharmacogenomics is a critical part of precision medicine in which genomic information is used to optimize the choice and dosing of medications. We have been building the Pharmacogenomics Knowledgebase (PharmGKB) since 2000. It is a repository of information about how human genetic information impacts drug response. Central to our effort is careful human curation of information about the interaction between genes, drugs, and diseases. However, as the literature has accelerated, there is considerable pressure on our curators to be efficient, and we have turned to methods of automatic extraction and analysis of text (natural langauge processing, NLP). In this talk, I will share a few vignettes of our work extracting relations from published text, and using it for discovery and to accelerate the curation processes at PharmGKB.