A breakthrough in artificial intelligence is set to revolutionize the early stages of drug discovery. Australian researchers, led by Monash University, have unveiled an innovative AI tool named PSICHIC (PhySIcoCHemICal) that promises to transform the process of identifying potential new medicines by streamlining virtual screening. This development marks a significant step forward in the field, addressing a crucial gap in predicting the strength of interactions between molecules and proteins, which is essential for drug discovery.
Computational methods in drug discovery are well-established, but they have often struggled with the efficiency and cost-effectiveness of predicting protein-molecule interactions. PSICHIC, however, offers a novel approach that combines sequence data with artificial intelligence to decode these interactions with remarkable accuracy. Published in the prestigious journal Nature Machine Intelligence, the study illustrates how PSICHIC eliminates the need for costly and less accurate processes such as 3D structural analysis, which has been a significant barrier in the field.
Dr. Lauren May, co-lead author from the Monash Institute of Pharmaceutical Sciences (MIPS), highlighted the tool’s capabilities, stating that PSICHIC has already shown its effectiveness in screening new drug candidates and performing selectivity profiling. In one notable example, the tool was used to compare experimental and AI predictions for a large compound library against the A1 receptor, a potential therapeutic target for numerous diseases. The results demonstrated that PSICHIC could efficiently identify a novel drug candidate and distinguish the compound’s functional effects, which is crucial for understanding how a drug might interact within the body.
Dr. May emphasized the transformative potential of AI in drug discovery, foreseeing a future where tools like PSICHIC significantly enhance the understanding of protein-molecule interactions. Data scientist and AI expert Professor Geoff Webb, from Monash’s Department of Data Science and Artificial Intelligence, noted that while other methods exist, they are often expensive and less accurate in predicting a drug’s functional effects. PSICHIC, on the other hand, bypasses these limitations by analyzing thousands of protein-molecule interactions without the need for high-resolution 3D structures.
Professor Webb explained that PSICHIC identifies the unique “fingerprints” of specific protein-molecule interactions through AI analysis, resulting in faster and more effective screening of drug compounds. This innovation not only speeds up the discovery process but also reduces costs, making it a more accessible option for researchers and pharmaceutical companies.
Dr. Anh Nguyen, another co-lead author from MIPS with expertise in AI approaches to drug-receptor interactions, stressed the importance of understanding these interactions. Drugs exert their effects by selectively interacting with specific proteins, and accurate prediction of these interactions is fundamental to drug discovery. Global efforts have been made to develop AI-based methods to determine how molecules might behave when interacting with protein targets, and PSICHIC represents a significant advancement in this area.
PhD candidate Huan Yee Koh, the first author from Monash’s Faculty of Information Technology, discussed the motivation behind designing PSICHIC. Koh highlighted that while AI has the potential to improve the robustness, efficiency, and cost of drug discovery, many existing AI systems suffer from over-reliance on pattern matching. This can lead to memorizing known patterns rather than learning the underlying mechanisms of protein-ligand interactions, ultimately hindering the discovery of novel drugs. PSICHIC addresses this issue by incorporating physicochemical constraints into its AI model, allowing it to decode these mechanisms directly from sequence data, bypassing the need for costly structures.
Professor Shirui Pan, co-lead author and an ARC Future Fellow with the School of Information and Communication Technology at Griffith University, praised PSICHIC for its accessibility. The tool requires only sequence data for operation, making it uniquely accessible compared to previous deep sequence-based methods. This approach provides a more accurate representation of protein-molecule interactions, bridging the performance gap between sequence-based methods and those relying on complex or structure-based methods.
To foster further research and collaboration, the PSICHIC team has made their data, code, and optimized model available to the broader scientific community. Interested researchers can access these resources at www.psichicserver.com, where they can explore the full study titled “Physicochemical graph neural network for learning protein-ligand interaction fingerprints from sequence data.”
The team behind PSICHIC, including co-lead authors Professor Geoff Webb and Dr. Lauren May, are available for interviews to discuss their groundbreaking work and its implications for the future of drug discovery. As the field continues to evolve, innovations like PSICHIC offer promising solutions to some of the most pressing challenges in developing new medicines, potentially accelerating the journey from initial discovery to clinical application. This tool not only enhances the efficiency of drug screening but also ensures a more reliable and cost-effective approach, paving the way for more rapid advancements in the pharmaceutical industry.
