Research section Research projects and references

Research: FORCE-FIELD-PROTEIN Folding Project #18239

Project #18239 overview

Project Summary AI Beta

This project uses computer simulations to study how proteins move and fold. They're testing different 'force fields', which are like rules for how atoms interact, using a well-studied protein called T4 lysozyme. The goal is to improve the accuracy of these simulations, helping scientists better understand how proteins work.
Automated summary; simplified and may not be fully accurate.

Project team

Manager(s)
Justin Miller
Institution
University of Pennsylvania

Work unit

Atoms
117,230
Core
0x24
Status
Public
Source material

Official Project Description

Force fields aren't only a thing in far off galaxies, but are also an integral part of molecular dynamics simulations.

Principally, molecular dynamics simulations are evaluating Newton's laws of motion iteratively.

Each atom in the simulation is given a position, velocity, and has some forces acting upon it.

We then take a short step forward in time (often 2-4 femtoseconds), update the positions of each atom based on the last known position, velocity, and acceleration, before re-evaluating the forces acting upon each atom.

Repeating this millions to trillions of times (or more), gives us a physics-based movie of atoms moving which we use to give insight into the behavior of our favorite proteins. One of the fundamental steps of this process is calculating the forces on each atom.

The collective model describing how to calculate these forces is called a force field.

Through the years, many force fields have been derived and refined, each one focusing on improving certain forces or behaviors of the simulation.

While tests are usually performed when force fields are redeveloped, it is difficult to achieve robust sampling (e.g.

many observations of rare events).

Here, we are continuing our efforts to catalog the performance and accuracy of these force fields.

In this project series, we use the well studied protein, T4 Lysozyme, as our test model.

Lysozyme is an antibacterial protein which destroys bacterial cell walls.

Lysozyme is an ideal system to use for evaluating force fields as many biophysical measurements have been performed on the system and several rare conformations (folds) of the protein have been observed.

We expect that our findings in this project series, along with the similar project series 18227-18230 and 18250-18255, will provide a strong benchmark to improve the accuracy in simulations, both on Folding@home as well as in the broader scientific community, to come. We are testing the following force field/water combinations in this project series.

If you are particularly excited about additional force field/water combinations, please reach out. 18235- Amber03 with TIP3P water 18236- Amber14sb with TIP3P water 18237- Amber19sb with OPC water 18238- Charmm36m with TIP3P water 18239- Amber19sb with OPC3 water 18240- Amber99SB-disp with TIP4PD-1.6 water 18241-Amber19sb with OPC3-pol water 18242-Amber99SB-star-ILDN with TIP4PD water.

Performance data

Hardware Performance for Project 18239

Compare community-sampled Folding@Home output for the GPUs and CPUs processing this project.

Data as of Sunday, 02 August 2026 21:53:05

GPU PPD Averages

Rank
Project
Model Name
Folding@Home Identifier
Make
Brand
GPU
Model
PPD
Average
Points WU
Average
WUs Day
Average
WU Time
Average
1 GeForce RTX 4060 Ti
AD106 [GeForce RTX 4060 Ti]
Nvidia AD106 10,027,048 1,024,331 9.79 2 hrs 27 mins