It is well-known that proteins are the cornerstone of life and play a key role in all biological processes. Therefore, understanding how they interact with the environment is critical to developing effective treatments and designing the basis for artificial cells.
Recently, the Protein Design and Immune Engineering Laboratory (LPDI) of the Institute of Bioengineering at the Swiss Federal Institute of Technology (EPFL), in collaboration with the Institute of Computational Science at the USI School of Information, Imperial College, and other units in the United Kingdom, has developed a groundbreaking machine learning-driven technique for predicting interactions between proteins and the environment and achieving a description of the biochemical activity of proteins based only on surfaces. In addition to deepening our understanding of protein function, this approach, known as MaSIF, can also support the future development of protein-based components in artificial cells. The study was published in the December 9 issue of Nature Methods.
In this new study, the research team took a large amount of protein surface data and entered these chemical and geometric properties into a machine learning algorithm and trained them to match them to specific behavioral patterns and biochemical activities. Then, they used the remaining data to test the algorithm.
“By scanning the surface of a protein, our method can define a ‘fingerprint’ that can then be compared between proteins,” said the leadauthor, Dr. Pablo Gainza of EPFL Bioengineering Institute and Swiss Institute of Bioinformatics.
Scientists have developed a new method to predict interactions between proteins and other proteins and biomolecules, as well as to predict their biochemical activity by looking only at their surfaces.
The team found that proteins with similar interactions all share a common “fingerprint”.
“The algorithm can analyze billions of protein surfaces per second,” said Bruno Correia, Ph.D., director of LPDI and corresponding author of the study. “Our research has important implications for the design of artificial proteins that allow us to program proteins to behave in a specific way simply by changing their surface chemistry and geometry.”
The method is published as open source and can also be used to analyze the surface structure of other types of molecules.