1 of 1

AMBER QM/MM Interface Development

Prathyush Katukojwala, Pietro Sette, and M. Evan Wildenhain

Supervised by Dr. Andreas W. Götz

Results

Our work involved two main aspects:

  • Extension of existing and development of new Fortran modules to couple sander to external QM programs.
  • QM/MM MD simulations to test the numerical accuracy and efficiency of our implementations.

To begin development, we learned Fortran as required for working on the QM interface modules. We added functionality to AMBER’s interface for 3 different QM programs. For ADF we added support for Becke integration grids and the ZLM electron density fit. For Q-Chem, which only supported QM calculations, we added support for QM/MM as well as various QM methods (MP2, DFT). We also increased the speed of QM calculations by implementing the use of restart files to accelerate SCF convergence at each MD time step. We added validation tests for both Q-Chem and ADF to for automatic testing when installing the software. Enhancements for the PSI4 QM code are underway.

To test and benchmark our additions to AMBER, we learned how to run MD simulations with the MD program sander. This involved compiling sander and learning its input and output formats. We had access to the supercomputers Gordon, Trestles, and Stampede and learned how to submit simulations via the queuing systems. The figures below show energy conservation and performance of QM/MM simulations with ADF using different numerical settings.

A.W. Götz, M.A. Clark, R.C. Walker, J. Comput. Chem. 35, 95-108 (2014).

D.A. Case, V. Babin, J.T. Berryman, R.M. Betz, Q. Cai, D.S. Cerutti, T.E. Cheatham, III, T.A. Darden, R.E. Duke, H. Gohlke, A.W. Goetz, S. Gusarov, N. Homeyer, P. Janowski, J. Kaus, I. Kolossváry, A. Kovalenko, T.S. Lee, S. LeGrand, T. Luchko, R. Luo, B. Madej, K.M. Merz, F. Paesani, D.R. Roe, A. Roitberg, C. Sagui, R. Salomon-Ferrer, G. Seabra, C.L. Simmerling, W. Smith, J. Swails, R.C. Walker, J. Wang, R.M. Wolf, X. Wu and P.A. Kollman (2014), AMBER 14, University of California, San Francisco.

AMBER (Assisted Model Building with Energy Refinement) is a molecular dynamics (MD) simulation software package to perform simulations in contexts ranging from drug design to artificial photosynthesis. MD simulations can use different levels of approximations:

  • Quantum Mechanics (QM) - high accuracy and ability to describe reactive events; computationally expensive
  • Molecular Mechanics (MM) - classical mechanical potential parameterized for equilibrium properties; computationally very efficient
  • QM/MM - uses QM potential for a region of interest (for example, enzyme active site), coupled with an MM description of the environment

AMBER supports QM/MM natively, but can be used with other QM packages to access an array of advanced QM approaches. The aim of this project is focused on enhancing the functionality and improving performance of QM/MM simulations with AMBER. Specifically, our primary objectives were the following:

  • Widen the range of QM programs accessible by sander
  • Increase functionality of QM methods accessible via these programs
  • Optimize default settings for numerically stable QM/MM simulations

Our task was to develop and test interfaces between various external QM programs (ADF, Q-Chem, PSI4) and sander, the main molecular dynamics simulation program in AMBER.

Objectives

References

We enhanced the QM/MM functionality of the AMBER MD software package to support additional QM programs and methods. Specifically, we added functionality for ADF and added support for Q-Chem to use MP2, density functionality theory and electrostatic embedding. Also, we reduced the computational cost by using Q-Chem’s restart files to improve the initial “guess” for SCF energy convergence. We added validation tests for ADF and Q-Chem to AMBER that can be run in an automated fashion.

We used the supercomputers Gordon, Trestles, and Stampede to test the numerical accuracy as well as computational performance of our implementations.

In working toward this project’s goals, we used research allocations on the Gordon and Trestles supercomputers of SDSC as well as Stampede of Texas Advanced Computing Center, learning to use the queuing system to run jobs of up to 48 hours as well as SSH into interactive shells. We also had the opportunity to gain familiarity with emacs, git, and Bourne shell scripting as part of this project’s development and testing work. Development of extensions to AMBER’s QM interfaces required us to learn how to program in Fortran.

This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1053575. Computer time was provided by SDSC and TACC via XSEDE award TG-CHE130083. AWG acknowledges financial support from DOE SciDAC (DE-AC36-99G0-10337) and NIH (R01 GM100934).

Acknowledgements

Introduction

Summary

Method

QM/MM MD (NMA in droplet of 508 SPC/Fw water molecules) energy conservation and performance with ADF (BLYP/DZP) using various integration grid and fit type settings.

QM/MM Simulation of NMA in Water