Use GROMACS Online

GROMACS: High-Performance Molecular Dynamics Simulation of Biomolecules from Laptops to Supercomputers

GROMACS is one of the most widely used open-source, free software codes in chemistry, used primarily for molecular dynamics (MD) simulations of biomolecules. The version 5 paper by Mark James Abraham, Teemu Murtola, Roland Schulz, Szilard Pall, Jeremy C. Smith, Berk Hess and Erik Lindahl (KTH Royal Institute of Technology, Oak Ridge National Laboratory, University of Tennessee and Stockholm University) describes how it reaches new performance heights through parallelization at every level. The code is distributed under the LGPL v2.1, and the authors note that the free license permits commercial use.

MD gives spatial and temporal resolution that experiments cannot. According to the paper, simulations with better force fields now sample motions on the microsecond scale, and ensemble techniques make millisecond-scale processes such as protein folding accessible. A typical user picks an initial configuration, describes the atomic interactions, runs a simulation and analyses the trajectory, which means evaluating millions of interactions per step. Making that fast on hardware ranging from a laptop to a supercomputer is a hard software engineering problem, and it is the focus of GROMACS development.

Why GROMACS Matters for Biomolecular Simulation

The paper emphasizes that computing resources are scarce, so minimizing resource use per simulation makes it possible to run more simulations, for example through ensemble methods. GROMACS aims for the highest absolute performance and efficiency on any hardware, and it runs on every architecture in the Top500 list as well as on embedded systems and laptops. Because MD problems have a fixed size, such as a 30,000-atom protein system, strong scaling and time to solution matter more than weak scaling.

How GROMACS Works: Multi-Level Parallelism

GROMACS 5 distributes work across several levels at once:

1. SIMD registers inside cores: a SIMD abstraction module lets non-bonded kernels be written independently of SIMD width, supporting 13 architectures.

2. OpenMP multithreading: several cores work on a single domain of atoms, which improves efficiency and extends the strong-scaling limit.

3. Heterogeneous CPU-GPU acceleration: short-range non-bonded work is offloaded to GPUs using a cluster-based Verlet pair-list scheme with automatic buffering, while the CPU keeps bonded forces and PME.

4. 3D domain decomposition: an eighth-shell scheme with dynamic load balancing in all three dimensions, with long-range electrostatics handled by particle-mesh Ewald (PME) on dedicated ranks.

5. Ensemble-level parallelism: built-in replica exchange, plus the separate Copernicus framework for free-energy calculations, Markov state modeling and the string method across many simulations.

All systems use triclinic unit cells internally, so shapes such as the rhombic dodecahedron are supported everywhere, and the authors state this can improve performance by up to 40% compared with a rectangular box for the same water thickness.

Simulation Capabilities

• Integrators: leap-frog and velocity Verlet, Brownian and stochastic dynamics, plus energy minimization, normal-mode analysis and simulated annealing.

• Force fields: 15 flavors of AMBER, CHARMM, GROMOS and OPLS are validated and included, with coarse-grained and implicit-solvent options.

• Advanced sampling: replica exchange, pulling and umbrella sampling, alchemical free-energy transformations, essential dynamics, Hamiltonian replica exchange and expanded-ensemble methods.

• Lennard-Jones PME: a new implementation that the authors report costs only 10-20% more than short cutoffs.

• TNG trajectories: a compressed trajectory format that improves on XTC compression.

Reported Performance and Scaling Results

The authors benchmark two membrane protein systems with PME and 1.0 nm initial cutoffs, as summarized below.

The authors expect accelerated clusters to deliver about 3 times the performance of CPU-only clusters with similar CPUs and K40 GPUs.

What is Tamarind Bio?

Tamarind Bio is a no-code bioinformatics platform built to give life scientists and researchers access to powerful computational tools. Many cutting-edge machine learning models are hard to deploy and use. Tamarind provides an intuitive, web-based environment that removes the complexity of high-performance computing, software dependencies and command-line interfaces.

The platform is designed for biologists, chemists and other researchers who may not have a background in programming or cloud infrastructure but want to run models on their own data. Key features include:

• A user-friendly graphical interface for setting up and launching experiments

• A robust API for integration into existing research pipelines

• An automated system for managing and scaling computational resources

Tamarind treats information and data security as a top priority, as detailed in its Trust Center and Terms of Service.

Accelerating Discovery with GROMACS on Tamarind Bio

• Biomolecular dynamics: simulate proteins, membranes and other biomolecules, the primary use the authors describe.

• Free-energy studies: apply the alchemical free-energy transformations and umbrella sampling that the paper lists.

• Enhanced sampling: use replica exchange and related ensemble methods to explore slow processes such as folding.

• Coarse-grained modeling: the paper notes that force fields such as MARTINI use the GROMACS infrastructure.

How to Use GROMACS on Tamarind Bio

1. Open the tool: log in to tamarind.bio and select GROMACS.

2. Provide the system: upload your structure and topology or prepared input files, as the form defines. The paper notes that online services can generate coordinates, parameters and topologies.

3. Choose the force field and model: select among the supported AMBER, CHARMM, GROMOS or OPLS force fields and a solvent model, if exposed.

4. Set simulation parameters: choose options such as integrator, temperature and pressure control, and run length, if exposed.

5. Run the simulation: submit the job and wait for completion.

6. Download the outputs: retrieve trajectories, energies and logs.

7. Analyse: use GROMACS analysis tools or your own scripts to study the trajectory.

Things to Keep in Mind

• The paper describes GROMACS version 5; features of later releases are not covered.

• The authors note that performance is best with balanced hardware, and that accelerators sit idle once forces are computed while constraints are enforced on the CPU.

• Because default mixed precision is used, a few kinds of simulations that require double precision need that mode specifically.

• The reported benchmarks used specific hardware and membrane protein systems, so throughput on other systems will differ.

Source: Abraham MJ, Murtola T, Schulz R, Pall S, Smith JC, Hess B, Lindahl E. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1-2, 19-25 (2015). DOI: 10.1016/j.softx.2015.06.001. Code: https://github.com/ElsevierSoftwareX/SOFTX-D-15-00003.