In the past decade, rapid synthesis and High Throughput Screening (HTS) of large libraries of compounds have resulted in significant numbers of molecules tested with individual receptors. As a public bioassay data depository portal, PubChem provides the largest primary HTS dataset obtained from various toxicity bioassays. However, extracting useful information from all these data for a large chemical library of specific interest becomes difficult. As a response to this challenge, we developed an automatic data mining tool to extract relevant bioassay data for animal acute toxicity. About 11,000 compounds with their experimental rat oral acute toxicity results have been compiled as our in-house animal toxicity database. We aimed to generate a set of bioassays which are relevant to animal acute toxicity. To this end, we developed an automated data annotation tool, which can perform natural language processing (NLP) to extract key information from the original bioassay profiles. After removing the untested compounds and those bioassays with no active responses, the first curation effort resulted in a comprehensive bioassay databases including 3,322 compounds tested against 555 bioassays. The second step is to use a stepwise validation procedure to gradually remove the irrelevant toxicity bioassays to our target animal acute toxicity. The bioassay response space resulted from this procedure consists of the top 50 PubChem bioassays. Most of these bioassays are used to test the chemical growth inhibition of cancer cells, so the relationship between them and acute toxicity is essential. We wish the method developed in this study could be used to understand the relationship between the current existing HTS data and complex bioactivities (e.g. animal toxicity). Through this way, we could avoid to re-test those compounds that have been tested by relevant bioassays and use this information to guide the development of alternative bioassay testing panel for animal toxicity.

Virtual profiling animal toxicants by automatic mining public bioassay data

246th National Meeting (2013)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.