GRTOOL: a GPU-accelerated utility to calculate atomic RMSD of biomolecular structures via molecular dynamics

Chemistry, in particular structural biochemistry, has directly benefited from the use of molecular dynamics (MD) simulations. This computational methodology significantly accelerates experimental research. One of the key techniques to the analysis of MD simulations is the calculation of the root mean squared deviation (RMSD). This technique is essential for assessing the stability of any MD trajectory as well as evaluating conformational changes throughout the simulation. For long simulations this primary calculation becomes costly, and an efficient method is needed. Currently, the use of graphics processing units (GPU) offer the best hardware platform to improve performance. In this work, the computational analysis program GRTOOL was developed to reduce the overall runtime as compared to currently existing tools. In GRTOOL, the use of parallel threads to overlap tasks combined with the GPU’s ability to process hundreds of atoms simultaneously resulted in a pronounced runtime reduction. A total of 11 unique parameters incorporated in the source code generated over 1,000 unique configurations to test for the best performance. The optimal configuration in GRTOOL lead to a 7x speedup of overall runtime, and over 43x speedup for the RMSD calculation alone. As MD trajectories become longer, such a tool will become increasingly necessary for timely analysis of biomolecular structural data. Furthermore, GRTOOL additionally provides a foundation for more advanced structural analysis on the GPU.