Australian researchers from Monash University have created a new artificial intelligence (AI) tool called PSICHIC (PhySIcoCHemICal) that aims to transform how scientists find new medicines. This tool is designed to make the process of virtual screening in early drug discovery faster, more reliable, and less expensive.

Using just sequencing data and artificial intelligence, co-lead author Dr. Lauren May of the Monash Institute of Pharmaceutical Sciences (MIPS) says PSICHIC can forecast how drugs and proteins interact. This eliminates the need for costly and erroneous methods based on 3D structures. "PSICHIC can essentially screen new drug candidates and profile their selectivity," the researcher said. "It can point out a new drug candidate and determine how the medication might affect our bodies."

Lead author Professor Geoff Webb of Monash's Department of Data Science and Artificial Intelligence highlights the advantages of PSICHIC. "Other methods for estimating protein-molecule interactions are costly and generally unreliable in projecting the effects of a medication. PSICHIC removes the need for three-dimensional structures, therefore accelerating and improving the process.

Another co-lead author from MIPS, Dr. Anh Nguyen, stresses the need to know how proteins and chemicals interact to generate effective drugs. "Global efforts to create AI methods for precisely determining these interactions are significant because they are essential for making medicines." Dr. Nguyen says.

Huan Yee Koh, a Monash PhD student, talks about how incorporating physicochemical constraints into PSICHIC's AI models improve efficiency and dependability of drug development. Mr. Koh explains, "This approach helps PSICHIC decodes the mechanisms of protein-ligand interactions directly from sequence data, making drug discovery more efficient and reliable."

Co-lead author Professor Shirui Pan of Griffith University notes that PSICHIC's unique accessibility stems from its dependence on sequencing data. "This approach offers a more accurate portrayal of protein-molecule interactions than past techniques," the researcher said.

Published in Nature Machine Intelligence, the work shows how PSICHIC could revolutionise virtual screening and advance knowledge of protein-molecule interactions, therefore opening the route for faster and more successful drug discovery.