Rashmi Kumari, Andrew M Lynn
School of Computational and Integrative Sciences, Jawaharlal Nehru University, New Delhi, India.
In the recent years, MM/PBSA combined with the MD simulations emerged as the fast and approximate method for the calculation of binding free energy (1-3). This approach is employed in the rescoring of the docked complex to remove false positives obtained from the docking methods (4-8). MM/PBSA is already implemented, optimized and validated in the AMBER package. MD simulations through AMBER are still computationally expansive and high throughput rescoring of the docked complexes is time consuming. Also, MD simulation module of AMBER package is not available as an open source program. These two major drawbacks motivated this work with the objectives of implementation, validation and optimization of MM/PBSA through open source programs. GROMACS and APBS are used for the MD simulations and PB calculation respectively. Whole method is implemented through Perl script and validated on large dataset having broad K i range. HIV protease I inhibitor complexes were taken as dataset and subjected to the 10 ns of MD simulations for sampling. MM/PBSA was used in the prediction of the relative binding free energy. After optimization of this method, a good correlation coefficient is obtained between calculated relative binding energy and the experimental logK i values. The optimized method is used to re-score the docked complexes of DHDP (DiHydroDiPicolinate) reductase, a key enzyme of Diaminopimelate pathway of M. tuberculosis. Docked complexes of some other known M. tuberculosis target are also re-scored which clearly discriminate between active and in-active compounds from docking hit-list. These re-scoring demonstrate that predicted activity by the MM/PBSA is different from the docked score and can be taken further for experimental testing. Read more…