AI helps to design new drugs

According to a press release, two scientists from the Swiss Federal Institute of Technology Lausanne (EPFL) and the University of Lugano (USI) have developed an innovative AI system to analyze proteins. Their surface structure can be used to predict how the building blocks of life behave. This is fundamental in designing new drugs.

This system is called MaSIF (Molecular Surface Interaction Fingerprinting). It was developed by Bruno Correia, Tenure Track Assistant Professor at the EPFL Laboratory of Protein Design & Immunoengineering (LPDI), together with Michael Bronstein, Professor of Computing at the USI and Imperial College London. Their work brought them to the front cover of the scientific journal “Nature Methods” in February.

They taught the system how to link the geometric and chemical properties of proteins to their interaction capability. After training, the algorithm is able to analyze billions of protein surfaces per second. Bronstein explains: “It is like a three-dimensional puzzle that can now be solved up to 10,000 times faster than before.”

The system developed by Bronstein and Correia will help to understand how the estimated 20,000 proteins in the human body interact. Moreover, it may also be used to develop new cancer drugs, so researchers could for example create molecules that bind to the protein complexes that disguise cancer cells against the immune system and inhibit them. Proteins can additionally be used to speed up reactions in industrial processes.