My main interests are to develop and improve tools for
biologists to discover reagents for their targets and
to use those tools for discovery projects.
My work can be divided into four areas:
- Target-based (docking) methods
- Ligand-based (2D) similarity methods
- Databases to support these approaches
- Collaborative research and other projects
Virtual Screening
Virtual screening is the most practical tool to leverage structure for ligand discovery. Yet many of the people who could make most use of it are not using it? Why? There continue to be substantial barriers to entry for non-specialists. My work focuses on making virtual screening easier to use and more reliable. This has lead us to develop DOCK Blaster, a free public-access virtual screening facility, based on DOCK molecular docking software.
- DOCK Blaster
- DOCK Software , created by Tack Kuntz and his group
Ligand-based approaches to ligand discovery
We have developed a technique that quantitatively groups and relates proteins based on the chemical similarity of their ligands. We call this the Similarity Ensemble Approach.
Databases
One of the barriers to entry to virtual screening has been the availability of a
database of compounds to dock. We therefore developed ZINC, a public-access database
of ready-to-dock compounds, which we give away for free to everyone from our
website.
To avoid bias in the assessment of virtual screening, decoys should be chosen to
resemble ligands physically, yet be chemically distinct. We therefore created DUD,
a directory of universal decoys, with 2950 ligands for 40 different targets. We give DUD
away for free to everyone.
We have created a database of high energy intermediates of metabolites, HEI, which we
have used to discover substrates of enzymes.
- ZINC - a public access database for virtual screening
- DUD - a directory of useful decoys
- HEI - high energy intermediates
- WINC -
Other research
I am also working to document this field, and collaborate on several other projects.
- Comparing HTS to vHTS (especially MLSMR)
- DISI, an effort to publicly document computational ligand discovery methods