Open Access News

I am writing in regard to the Aug. 23 open letter by American Chemical Society President William F. Carroll about discussions between ACS leadership and NIH staff regarding PubChem....PubChem records do not systematically include experimentally derived properties, information about suppliers, information about synthesis and reactions, links to patents, or links to much of the chemical literature. Both ACS leadership and NIH staff led by NIH Director Elias A. Zerhouni have worked hard to find ways to avoid unnecessary duplication between PubChem and private-sector databases such as those provided by Chemical Abstracts Service. As a chemist and the director of an institute that supports a large amount of chemical research, I have personally participated in essentially all of these meetings. Unfortunately, Carroll's letter misrepresents several important aspects of these discussions.

First, while PubChem is focused on biomedically relevant compounds, neither ACS nor NIH has been able to develop a suitable definition of biomedical relevance that would accurately guide what compounds should or should not be included in PubChem. And contrary to a statement in Carroll's letter, no assurances about the number of compounds to be housed in PubChem have been given. Second, the expansion of PubChem from 600,000 compounds to more than 3 million compounds noted in the open letter is due largely to the inclusion of the ZINC database (C&EN, Feb. 14, page 8). This database was developed with support from a grant from NIH (R01GM071896) for the explicit purpose of making publicly available a large and diverse set of compounds for computational screening against biomedical targets. The NIH Roadmap Molecular Libraries Initiative explicitly includes support for developing and testing "new algorithms for computational chemistry and virtual screening." Thus, Carroll's statement that "most [of the newly posted compounds] have no particular biomedical focus" is inaccurate and misleading. Third, while we appreciate the offer from ACS to build a PubChem-like database for free, the statement that this would result in significant cost savings to NIH is inaccurate. The great majority of the expense related to the development of PubChem involves the thorough integration of information regarding chemical entities with the biomedical literature in PubMed, protein structural information, biological activities, and other data housed at NCBI. This integration is where the great value of PubChem for biomedical research lies....From our first discussions with ACS, NIH has offered to provide links from records in PubChem to records in CAS databases. This would allow researchers to navigate easily from the biomedically oriented information in PubChem to the extremely valuable chemically oriented data developed and maintained by CAS. Together, these and other databases will serve the research community as we apply chemical tools to understand biological systems and to improve human health.