Researchers create new computing tools to make more accurate predictions of protein structures

An algorithm was installed at the current location of the village forge, its powerful mathematical hammer hammering the proteins to shape it.

The blacksmith profession is a valid analogy for what scientists at Rice University have established: a new method of creating accurate structural models of proteins that uses far less computational power than existing approaches using brute force.

According to physicist Peter Wolynes of Rice, of the Center for Theoretical Biological Physics (CTBP) of the physicist, the purpose of the structural models produced by the computation is to be as detailed and useful as those produced by laborious experimental means , including X-ray crystallography, which provides detailed information. locations for each atom in a protein.

The new method inspires metallurgy. Like the blacksmith who must not only heat and cool a metal, but also hit it just to bring it closer to a useful product, the Rice project led by Wolynes and his former pupil, Xingcheng Lin, applies a force to the strategic points when simulation of protein models to speed up the calculation.

"A big question is whether we could ever become more confident in the accuracy of the results of a simulation than in the results of X-ray experiments," said Wolynes. "I'm about to say that's where we are now but, of course, time will tell."

The study appears this week in the Proceedings of the National Academy of SciencesResearchers have been using X-ray crystallography for more than a century to know the position of atoms in molecules from their structures in protein crystals. This information is the starting point for structural biology studies, and accuracy is considered essential for designing drugs that interact with specific proteins.

But crystalline structures provide only a snapshot of a protein that actually changes its overall shape and its detailed atomic positions as the protein performs its work in the cell.

Wolynes and his colleagues have long been the first to use computational methods to predict folded structures from the energetic landscape encoded in the amino acids of the protein. In the new work, they address the detailed placement of the amino acid side chains that can be pushed in this or that way by an algorithm starting from a moderate resolution view of the overall structure.

"To achieve the desired resolution, from the original coarse-grained models, we would normally need to run the computer for two months," he said. "But we discovered that we could first simulate the motions of the coarse-grained model to find the motions that would most substantially alter the binding patterns in the molecule.

"Some moves do not do anything at all: you may be fighting your hand, but the important thing is to bend your elbow," Wolynes said. "So we found a recipe for selecting the most significant motions and using them to bias another high-resolution simulation.We deliberately used the force to push the proteins just that way, and then we looked at the structures that to see if they were more stable than those we started with. "

Like a blacksmith hammering sand into a piece of metal, Rice's team also found methods to eliminate the "grain" of their patterns: slow, slow-moving side chains that slowly move and slow-absorbing dynamics computer time in the form of a folded protein. Delete the grain did not change the result, but made the calculation much faster.

"Metalworkers heat the parts and cool them for annealing, but they also find a way to make the big movements that will not happen spontaneously if you simply keep the metal at a high temperature," Wolynes said. "We have been annealing with coarse-grained models for a long time, but blacksmiths also hammer the metal to extract sand, or slag, and this has also caused us to mechanically deform proteins."

Wolynes said that CTBP has been methodically updating its protein folding and structural prediction models with new computer languages ​​over the years, allowing researchers to tackle More complex problems.

"Recoding the models allowed us to look at molecules ten times larger than before," he said. "There is no new physics, just new programming and better parallel computers, but that makes a real difference in the practical problems we can now solve."