Protein Folding in 2022

The protein folding problem has essentially been solved. First of our What's Hot in 2022 series.

Proteins are made up of long “strings” of amino acids joined together by peptide bonds. One of the great mysteries of science is embodied in the “ Levinthal paradox ”: given the great flexibility of these polypeptide sequences, the number of possible shapes (or “conformations”) of a protein is mind-bogglingly huge (Levinthal himself estimated 10300). How is it then that a protein can fold into the correct conformation to exert its biological function and do so on a milli- or micro-second timescale when to sample all the possibilities would take longer than the age of the universe?!

The resolution of this paradox is that a protein does not have to sample all possibilities, but rather that the folding is sped up and guided by the rapid formation of local interactions which then determine the further folding of the polypeptide chain. Insights such as this have informed the development of computational protein folding algorithms over recent decades. The motivation behind such efforts is that, if we were able to predict reliably the conformation of a protein, given just its amino acid sequence, this would have remarkable implications for biology in general and structure-based drug design in particular. No longer would we have to wait for the experimental determination of protein structures, which still sometimes elude us, despite advances in X-ray crystallography and, more recently, cryo electron microscopy. We would be able to design drug molecules using just the model generated by the computer!

Protein Folding Developments

Until recently, this seemed a distant hope towards which we were inching step by step. But then came AlphaFold2 , an AI-based protein folding program developed by DeepMind. We wrote briefly about its predecessor, AlphaFold, in 2020, but even then we could not foresee how quickly the method would develop and the astounding success it has achieved. Even the organisers of the protein folding competition CASP14 were forced to concede that the protein folding problem had essentially been solved!

We are only just beginning to see the implications of this incredible leap forward in technology. But thanks to the program’s code being made open source and to the Deepmind team releasing a large database of protein structures predicted by AlphaFold2, the impact will be felt fast and wide as 2022 progresses.

For a more detailed discussion of this topic, the reader is directed to this very recent Technology Feature published in the leading scientific journal Nature.

David E. Clark, PhD, Research Leader, Computer-Aided Drug Design

Our What's Hot series are annual forecasts provided by Charles River's thought leaders. Check out here what our scientists are predicting for 2022 and what they called out in years past.