Collaborative Drug Discovery, a software and database firm based in Burlingame, Calif., has augmented its online CDD Web 2.0 offering to include data from the Psychoactive Drug Screening Program Ki database at the University of North Carolina Medical Center in Chapel Hill, which includes 47,312 inhibitor equilibrium dissociation constants, or Ki values, for 699 receptor targets.
CDD said the PDSP Ki database is the largest open-access chemical sub-structure and similarity searchable G-Protein Coupled Receptor Ki database, and marks the company’s “first foray” into GPCRs, an area of central importance in commercial drug discovery.
Barry Bunin, president of CDD, said that the company added new functionality to the publicly available data in PDSP Ki, allowing users to mine it according to chemical and biological criteria.
The PDSP Ki database is part of a psychoactive drug screening center at UNC Chapel Hill led by Bryan Roth and funded by the National Institute of Mental Health. The lab performs functional and pharmacological screening of compounds, as well as Ki determinations, which measure a drug’s affinity for a given enzyme and indicate a drug candidate’s inhibitory potential.
The database includes more than 45,000 Ki values, or affinity values, for drugs and drug candidates at G-protein coupled receptors, ion channels, transporters, and enzymes. It is updated daily, and “gets between three-quarters of a million and a million hits a year, so it gets a lot of activity,” Roth told BioInform.
The PDSP Ki database is housed in the public domain section of the overall CDD database and joins 12 other data sources in the CDD system, which includes chemical and biological data for more than 40,000 compounds, 1,700 FDA-approved drugs, orphan drugs, compounds from public data sources, as well as more than 25,000 compounds available for purchase.
SMILE for Users
Roth said that the CDD has created “a very nice enhancement of our database” that makes the PDSP Ki data “a lot more useful” for researchers in many areas of drug discovery.
Previously, he said, “you could basically download the SMILES [Simplified Molecular Input Line Entry System] codes separately but it wasn’t really searchable in terms of chemical structure, to find related compounds, and then find their pharmacology.”
One focus of Roth’s research is G-protein coupled receptors, which more than 50 percent of approved drugs target. Among his research goals is exploring off-target effects of drugs and searching for novel compounds for the treatment of depression and other central nervous system disorders.
“The database gets between three-quarters of a million and a million hits a year, so it gets a lot of activity.”
Given its NIMH funding, the PDSP Ki database will remain a public and thus freely accessible database, Roth said. CDD “basically downloaded it, which anyone is free to do,” he said. The enhancements by CDD “will be very valuable to people who want to mine the data,” he added.
CDD’s Bunin said that the company’s enhancements now allow users to search the PDSP Ki data by chemical structure for the first time. This is an important functionality in drug discovery, he noted, as researchers explore biology and seek to mine structures with structural characteristics of their choosing.
Bunin did not elaborate on the computational details of the integration, but said “it wasn’t easy.” He and his team found “a way of automating the readouts in our database based on the architecture of the data,” he said.
CDD, founded in 2004, focuses on helping academic, foundation-based, and corporate researchers share data, Bunin said.
The company has integrated proprietary drug discovery data and public data to allow customers and users to archive and mine their data, search for chemically similar compounds, potency, and toxicity profiles. Data can be completely public, or “you can keep that data 100 percent private,” he said.
Bunin could not disclose all of CDD’s users, but said they include major universities such as Columbia University and several campuses in the University of California system, St. Jude Children’s Research Hospital, and companies such Asinex, Melior Discovery, Semafore Pharmaceuticals, and “premier ones that we can’t mention.”
Rule of Fiving
Chemist Christopher Lipinski, former chemist at Pfizer and inventor of the “rule of five” to determine if a small molecule is likely to be an orally active drug, donated CDD’s first dataset, a list of all known drugs and their therapeutic indications. Lipinski also introduced Bunin to Roth.
“New datasets [are] coming in all the time,” Bunin said. The GPCR database with the Roth dataset is “a particularly interesting one,” he said, given the importance of GPCRs in commercial drug development.
He acknowledged that GPCR data can be accessed in other databases such as the KEGG and GLIDA databases hosted at Kyoto University, but rather than compare CDD’s offering to these systems, he recommended that users explore the company’s resources to judge ease-of-use and the value of mining the chemical and biological data.
When datasets are in an electronic format such as the PDSP Ki database, they are uploaded into the CDD database in batches and updated regularly. “Other examples have been more arduous,” he said, including the re-drawing of 12,000 chemical structures from a 2,000-page book before uploading them into the database.
His company has also developed a software platform to enable drug discovery collaboration. For example, Bunin said, a start-up, an academic lab, a contract research organization, and a large pharmaceutical company will want to securely share data including structures for substructure queries. “You can share your data or subsets of data anywhere around the world,” he said.
The firm has spent much effort on the graphical user interface and collaborative functionalities for screening centers or virtual start-ups that are outsourcing work, he said. “You can have a dashboard so you see what was done, which minute, by whom; you can send messages back and forth and datasets back and forth,” he said.
CDD’s focus on the pre-clinical drug discovery data domain and medicinal chemistry sets it apart from competing platforms, he said. “There’s a lot of domain expertise in the technology, the company, and the collaborators,” Bunin said. The company is currently hiring programmers with scientific software development experience.