Skip to main content
Premium Trial:

Request an Annual Quote

Young Investigator Profile: John Chodera

Assistant Professor, Memorial Sloan Kettering Cancer Center

Recommended by Vijay Pande, Stanford University

NEW YORK (GenomeWeb) – Right now, a lot of drug discovery is people finding new drugs nearly by accident.

In his lab at the Memorial Sloan Kettering Cancer Center, John Chodera is trying to change how people go about designing drugs. He wants to make the process more akin to how engineers build bridges, buildings, or airplanes. Engineers have computational models that tell them whether a bridge will stand, a building will sway with an earthquake, and whether an airplane will be aerodynamic.

"We don't have that quality of computational models at the microscopic scale, so that's what my lab is focused on trying to change," Chodera said.

In particular, his lab is using computational models to design small molecules to target the kinases that are often mutated, dysregulated, or otherwise activated in cancer. This, he noted, is tricky as there are some 500 human kinases, and they all look rather alike.

Once his model comes up with something interesting, his lab then tries to improve it experimentally. This, too, he plans to automate so that the computers come up with hypotheses, add experiments to the queue to test those ideas, and feed the resulting measurements and its analysis of them back into the model.

That way, he added, the quality of the model is continually being tweaked and he can work with collaborators to actually design the small molecules.

Right now, he and his team are working their way through the kinome and expressing the kinases they can in E. coli to search for inhibitors. And then in conjunction with Stanford University's Vijay Pande, Chodera's former postdoctoral advisor, they are using Folding@Home to characterize the catalytic domains of the kinome to yield a sort of atlas of the kinase conformations.

"By using that, we'll be able to both validate this [our findings] using the biophysical experiments in our laboratory and then use it as a key tool for designing new allosteric inhibitors that might be much more highly targeted than the current generation of inhibitors," he added.

Funding, of course, can be problematic. Chodera noted that his work can sometimes fall in the gap between what the National Science Foundation funds and what the National Institutes of Health funds.

"Trying to get funding for this kind of thing where we need to characterize the basic biophysics in a quantitative manner that allows us to improve the models to the point we can actually do this kind of biology and drug discovery is difficult," he said.

Paper of note

Drug discovery, he said, tends to follow fads, and one that crops up often is entropy-enthalpy compensation. This train of thought says that there's a point at which improving the enthalpic interactions between a ligand a receptor will affect the conformational entropy, so the binding affinity actually can't be improved any further. Some people, Chodera said, reason this may be a problem for drug discovery work.

But in a review paper that came out in the Annual Review of Biophysics in 2013, he and the University of California, Irvine's David Mobley argued that both entropy and enthalpy are difficult to measure and the issue doesn't pose much of a problem. Further, Chodera said it's not even really what researchers should be interested in, which is the binding affinity and that's easier to measure.

"The main message of that paper was just to try to dispel the things that aren't true behind this fad as well as get people to focus on what is actually important in drug discovery," Chodera said.

Looking ahead

In the future, there will need to be a more rational approach to drug discovery, Chodera said. Soon, he added, computers will be fast enough and models will be good enough so that drug researchers will have predictive models just as engineers do.

To do this, though, researchers will have to take a look at their failures, just as bridge engineers have, to learn from their mistakes and make models that are robust to those failures.

"These models will be the things that take us through to the next decade of drug discovery," he added.

This is the fourth in a series of Young Investigator Profiles for 2015 that will appear on GenomeWeb over the next few months.