Brian K Shoichet
University of California-San Francisco, CA·Pharmacology
Finding a molecule that binds a particular protein — the first step toward most drugs and chemical tools — is still largely a matter of trial and error. Computers can shortcut it by virtually fitting millions or billions of candidate molecules into a protein's binding pocket and ranking them, a technique called docking.
- Fields
Pharmacology
Chemistry
Biophysics
- Topics
molecular docking
structure-based drug design
virtual screening
- Grants
- 1 known grant, through Aug 2027
What they work on
- Test whether docking performance keeps improving as virtual libraries grow into the tens of billions to trillions of molecules, or saturates, and whether such libraries drift away from biology-like chemical space in ways that degrade hit quality.
- Develop faster search strategies, including look-ahead pattern matching, and improve the physics of docking by treating flexible receptors and the energetic penalties of ligand conformational and desolvation strain in simplified model cavity sites.
- Predict protein targets from a molecule's chemical structure using chemoinformatics, asking whether widely consumed nutraceuticals have specific targets and using network pharmacology to find compounds modulating human proteins subverted by SARS-CoV-2.
- Apply ultra-large library docking to biologically important receptors: bespoke libraries for aminergic GPCRs, extension to membrane transporters, and chemical probes for the yeast mating-factor receptor Ste2, where yeast genetics can validate hits.
- Determine how colloidal aggregation — compounds clumping and sequestering proteins — arises and distorts early screening, and test whether such aggregates can be repurposed for drug delivery.
- Investigate phospholipidosis, the drug-induced buildup of lipids in cells, as a previously underappreciated confounder of cell-based ligand discovery.
More about the research
But the scoring that decides which molecules look good is approximate, and it is rarely clear which parts of it are failing. This research builds better computational methods for predicting what binds what, tests them in systems simple enough to reveal exactly where they go wrong, and then applies them to proteins of biological interest.
Why it matters
Faster, more reliable prediction of which small molecules bind which proteins lowers the cost and time of finding chemical probes and drug leads. The methods, model systems and insights into misleading artifacts such as colloidal aggregation and phospholipidosis are tools other chemical biologists and drug discovery scientists can use directly, and individual molecules discovered along the way could illuminate cell signaling or serve as starting points for new drugs.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Steven Van Lanen
University of Kentucky, KY·Pharmacology
Develop and run virtual screens — target-based, pharmacophore-based and machine-learning-based — to prioritize compounds for center projects before any laboratory testing.
James Solomon Fraser
University of California-San Francisco, CA·Biophysics
Improve computational methods for finding and modeling alternative protein shapes in cryoEM density maps, extending the multiconformer modeling the group built for X-ray crystallography.
Seiya Kitamura
Albert Einstein College of Medicine, NY·Chemistry
Couple the SuFEx click-chemistry high-throughput medicinal chemistry platform with covalent ligands, molecules that bond permanently to proteins, to develop chemical probes for individual underexplored proteins that lack distinct binding sites.
The school
University of California-San Francisco: 89 labs on PI Pages, $2.1B a year on research, 164 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.