The pharmacological and medicinal chemistry literature connects activity data reported for drugs and research compounds to their explicit chemical structures.  However, the traditional entombing of these relationships for millions of compounds among millions of PDFs is retrospectively acknowledged as massively problematic especially from text mining restrictions.   The challenges for extraction from the literature become massive as PubMed pips 23 million,  The situation is at least partially ameliorated by a range of resources that extract the entity and data relationships the authors put “in” to the PDFs back “out” into structured database records  The history of the public manual efforts going back at least a decade include the MeSH system, BindingDB,  Comparative Toxicology Databse, ChEMBL and  the IUPHAR/BPS Guide to PHARMACOLOGY (GtoPdb). These have been joined more recently by automated extraction of PubMed abstracts by IBM (in parallel with their patent extraction) and Springer who recently extracted chemistry from 2.5 million of their papers.   This presentation will  review these sources that are identifiable in PubChem and explain how GtoPdb establishes its own journal to PubChem to PubMed linkages. It will also a) compare the relative coverage and extraction specificity with other sources b) point out some of the caveats associate with this,  c)  literature/patent content overlaps and d) general challenges of connecting chemistry to the biology described in the literature corpus.

Reciprocal journal-to-chemistry connectivity in PubChem from the IUPHAR/BPS Guide to Pharmacology and other sources

256th National Meeting (2018)

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