Testing and improving alchemical techniques for predicting protein-ligand binding

University Of New Orleans, New Orleans LA

Investigators

Linked publications, trials & patents

Abstract

DESCRIPTION (provided by applicant): Pharmaceutical drug discovery costs a tremendous amount of time and money. Now roughly $1 billion goes into bringing each new drug to market, on average [1]. An important factor in this cost is the amount of trial and error involved in finding initial hits which modulate the function of a biomolecule, and then refining these into leads which have adequate affinity for the biomolecular target and other desirable properties. Ideally, computational methods would drastically reduce the amount of trial and error involved, suggesting hits in advance of experiment, and guiding the refinement process by predicting chemical changes to improve affinity while maintaining drug-like properties. Current computational methods are not reliable enough to facilitate this radical change in the drug discovery process. This proposal applies the PI's recent innovations in predicting binding affinities based on computer simulations to several different systems of increasing biological relevance, [and proposes further innovations in orientational, conformational, and sidechain sampling]. This will result in further improvements in the approach, [and tests of its accuracy], bringing it closer to the point where it will be applied in a drug discovery context. The proposed approach uses alchemical absolute free energy calculations based on molecular dynamics simulations. This is one of the most physically realistic approaches available, and one of the most promising in terms of accuracy. This project's aims are to (1) improve the accuracy of binding free energy techniques in a polar model binding site in T4 lysozyme, using alchemical techniques with new algorithms to overcome sampling problems; [(2) predict binding affinities of trypsin inhibitors in a blind test, following this up with substantial analysis and additional calculations; and (3) compute binding affinities of an extensive set of DNA gyrase inhibitors, testing against experimental data and rationalizing observed trends.] In each of these projects, we will compute absolute [or relative] binding affinities for each potential inhibitor studied, beginning from a set of possible binding modes generated by docking potential inhibitors into one existing unbound structure of the protein. Using this set of possible binding modes as starting points for simulation means that bound structures of the individual inhibitors do not need to be known in advance for accuracy. Ultimately, this work will pay off in improved methods for affinity calculation. These will be applied to problems in drug discovery once the accuracy, speed, and reliability are sufficiently high. This work will play a key role in achieving necessary levels of accuracy and reliability. PUBLIC HEALTH RELEVANCE: Pharmaceutical drug discovery produces dramatic public health benefits through new and improved treatments for common diseases and disorders, but it is an expensive, time-consuming process involving much trial and error, and failure is common. This proposal refines and applies computer simulation techniques to predicting association between proteins, life's molecular machines, and potential pharmaceutical agents. This work has the potential to drastically change the early stages of the drug discovery process, aiding the development of new drugs and paying public health rewards!