Snapshots of the six coarse-grained models of proteins, shown as 2D projections. Credit: arXiv (2025). DOI: 10.48550/arxiv.2501.02424

The chains of amino acids that make up proteins are critical to every form of life. The complex ways that these proteins fold and interact has fascinated researchers for decades. Exactly how a protein folds determines its function. For instance, a particular protein can take on the job of transporting molecules, attacking invading cells, or repairing DNA.

A significant amount of progress has been made in this field of research, but much is still unknown about the folding process. Now, Yale researchers have figured out how to create computer models that simply, but accurately represent these proteins. The results are published in Physical Review E.

Every protein has a specific function, and when it doesn’t fold how it should, diseases and other serious disorders can result. A better understanding of how proteins fold could boost the creation of new pharmaceuticals to treat misfolding disorders.

Current computational models that operate at atomic resolutions, however, cannot simulate the folding of most proteins; the complexity makes the calculations prohibitively long. Further, the Protein Data Bank, a public database of protein structures, has the known structures of only 40% of human proteins.

“Therefore, we currently do not know the true structures and folding pathways of 60% of proteins,” said Corey O’Hern, professor of mechanical engineering, who led the study.

The researchers set out to develop a computational model that captured, as simply as possible, the most important structural properties of folded proteins.

Computational models can have different resolutions, O’Hern explained. They can describe the atomic scale—that is, they can model every atom in the protein. Or the models can be “coarse-grained,” in which groups of atoms in the protein can be represented by a single unit. For instance, the researchers might use a single spherical object to model the backbone atoms of a given amino acid and another spherical object can represent the side chain atoms of the amino acid.

“We studied a range of computational models from atomistic to very coarse-grained to determine the coarsest resolution that was able to accurately capture the structural properties of folded proteins,” he said.

The researchers compared the predictions of computational models to the protein core density and other structural features from thousands of proteins. They then created coarse-grained computational models with simplified representations.

“The coarsest model had one spherical bead per amino acid, compared to the smallest amino acid, which has 10 atoms,” O’Hern said. “The model with the highest resolution would include all of the atoms in each amino acid. However, including all atoms of the protein in the model is not always necessary to capture key structural features of proteins.”

In doing so, the researchers reduced the computational complexity, allowing for simulations of protein folds not possible with all-atom models.

“With a coarse-grained protein model that quantitatively captures protein structure, we will be able to fold the 60% of proteins with unknown structures,” O’Hern said.

More information: Jack A. Logan et al, Effect of stereochemical constraints on the structural properties of folded proteins, Physical Review E (2025). DOI: 10.1103/9wf9-ywhw. On arXiv: DOI: 10.48550/arxiv.2501.02424

Citation: Simplified protein models enable simulations of unknown folding patterns (2025, November 7) retrieved 7 November 2025 from https://phys.org/news/2025-11-protein-enable-simulations-unknown-patterns.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.