Development Of Advanced Computer Hardware And Software

National Heart, Lung, And Blood Institute

Investigators

Linked publications, trials & patents

Abstract

Migration of CHARMM to GPUs Graphical Processing Units (GPUs) are modern processors with the ability to support the concurrent execution of thousands of threads at the same time. In this work we have redesigned the CHARMM codebase from a heterogeneous CPU-GPU architecture to a GPU-only architecture. Several features of CHARMM have been implemented/optimized to utilize the underlying processor threads efficiently. Another important feature of the new implementation is the focus on modularity of the code to support easy extension and adherence to the state of the art software development best practices. We implemented support for multiple PSFs which acts as a unifying method for a number of free energy methods. Improving the speed of volumetric density map generation via cubic spline interpolation The creation of a 3D grid of atomic density (i.e. a volumetric map) is one way to easily view the long-time average behavior of data generated from molecular dynamics simulations, particularly when trajectories reach millions of frames. Volumetric maps can be generated using a grid by approximating each atom with a Gaussian function centered on that atom and spread over neighboring grid cells. However, the calculation of the Gaussian function requires evaluation of the computationally costly exponential function. We have increased the speed of generating volumetric maps from molecular dynamics trajectory data by replacing the expensive exponential function evaluation with an approximation that uses interpolating cubic splines. This new method has been implemented in the CPPTRAJ molecular dynamics analysis program. Adding compression to NetCDF molecular dynamics trajectories Molecular dynamics (MD) simulations can now be simulated on large systems (> 100,000 atoms) for increasingly longer time scales (microseconds and beyond) due to recent advances in computer software and hardware, in particular because of the adaptation of MD software to graphics processing units. As the size and length scales of molecular dynamics simulations increase, so too does the size of the corresponding trajectories. Modern MD simulations often generate hundreds of gigabytes of data. As a result, there is great interest in being able to store these trajectories in as efficient a way as possible without sacrificing too much precision. We are currently working on adding a quantization-based compression scheme to the popular Amber NetCDF trajectory format, utilizing the underlying HDF5 framework to allow compression and decompression to be done on-the-fly. We are also exploring how quantization and compression affects the precision of not only atomic positions (as is typically done), but also the energies and forces. A tool for preparing PDB files for simulation with the Amber software package Setting up molecular dynamics simulations from experimentally-determined structures is often complicated by a variety of factors, particularly when the structure to be simulated contains carbohydrates (e.g. the SARS-CoV-2 spike protein), since these have several forms and be linked in a variety of ways. We are currently developing a stand-alone tool implemented in the widely-used and freely-available software CPPTRAJ that can be used to facilitate building structures for use with the Amber Biomolecular simulation package. This tool will parse a given PDB file and automatically identify carbohydrate forms, chirality, and linkages, and change residue/atom names accordingly for use with the GLYCAM force field. In addition to being a stand-alone program, this process requires no user intervention which differentiates it from existing web-based tools. The tool will also provide the necessary commands for bonding carbohydrates and creating any disulfide bonds. The tool will be customizable via user-editable text files, potentially allowing it to be used for force fields other than GLYCAM. Psi45 and GEM The Psi4 quantum chemistry code has been updated to perform fits to the Gaussian Electrostatic Model (GEM), allowing automatic parameterization for next-generation force field methods. Other recent developments include improved integral screening and implementation of the continuous fast multipole method, to increase efficiency. Polyrate enhancements The Polyrate software package sets a reference for the calculation of rate constants using variational transition state theory with multidimensional tunneling. We have contributed to the last version of the code by updating it and implementing new tunneling methods that allow to study proton transfer reaction. The new implementation will allow to update the interfaces of this code to the CHARMM software package and will allow us to carry out chemical reaction dynamics for macromolecules using the QM/MM methodology. LoBoS hardware expansion We purchased 21 machines based on a single AMD CPU and dual GPUs( 20 dual Nvidia A100 machines and 1 Dual AMD Radeon 6700XT machine) to improve our LoBoS cluster for performing fast molecular dynamics simulations.