Scientists on the University of Toronto’s Donelly Centre have developed a cutting-edge AI model called PepFlow that may predict the varied shapes that peptides adopt with unprecedented accuracy.
Peptides are small molecules made up of amino acids, the constructing blocks of proteins.
While peptides are much like proteins, they’re much smaller and more flexible, allowing them to fold right into a huge number of shapes.
A peptide’s specific shape is crucial since it determines the way it interacts with other molecules within the body, which in turn dictates its biological function.
Predicting the structures of proteins and peptides has been a longstanding challenge in biology. Due to the complex math involved, it’s a superb problem for machine learning.
In recent years, AI models like AlphaFold 2 and three, developed by Google’s DeepMind, have revolutionized protein structure prediction.
AlphaFold2 uses deep learning to predict a protein’s almost definitely 3D structure based on its amino acid sequence. But while AlphaFold2 has been incredibly successful for proteins, it has limitations in terms of highly flexible molecules like peptides.
“We haven’t been capable of model the complete range of conformations for peptides until now,” said Osama Abdin, the study’s first writer.
Pepflow, documented in a study published in Nature Machine Intelligence, “leverages deep-learning to capture the precise and accurate conformations of a peptide inside minutes.”
PepFlow employs AI models inspired by Boltzmann generators. These models learn the elemental physical principles that govern how a peptide’s chemical structure determines its spectrum of possible shapes.
This allows PepFlow to accurately predict the structures of peptides with unusual features, similar to circular peptides formed through macrocyclization. Macrocyclic peptides are particularly interesting for drug development as a consequence of their unique binding properties.
What sets PepFlow aside from models like AlphaFold2 is its ability to predict not only one structure, but the whole “energy landscape” of a peptide.
The energy landscape represents all of the possible shapes a peptide can take and the way it transitions between these different conformations.
Capturing this structural complexity is vital to understanding how peptides function in numerous biological contexts.
PepFlow can inform drug development through the design of peptides that act as binders. #DrugDiscovery
Learn more 👉 https://t.co/eAKOg5e7Cz pic.twitter.com/mYP9YeiCOe
Significance
The ability to predict highly accurate peptide structures has major implications for developing peptide-based therapeutics.
“Peptides were the main focus of the PepFlow model because they’re very necessary biological molecules they usually are naturally very dynamic, so we want to model their different conformations to know their function,” explained Philip M. Kim, the study’s lead investigator.
“They’re also necessary as therapeutics, as could be seen by the GLP1 analogues, like Ozempic, used to treat diabetes and obesity.”
Peptide drugs have several benefits over traditional small-molecule drugs and bigger protein-based therapeutics. They’re more specific of their actions, have lower toxicity than small-molecule drugs, and are cheaper and easier to supply than larger protein drugs.
PepFlow could speed up the invention and development of recent peptide-based medicines by enabling the design of peptides with therapeutic properties.
“It took two-and-a-half years to develop PepFlow and one month to coach it, nevertheless it was worthwhile to maneuver to the following frontier, beyond models that only predict one structure of a peptide,” concluded Abdin.
This follows the discharge of EvolutionaryScale ESM3 this week, a frontier generative model for biology, which also focuses on proteins.
