Harvard's new AI improves protein analysis, leading to safer medications
Harvard Medical School (HMS) researchers have developed an innovative artificial intelligence (AI) tool, named PINNACLE, designed to understand the behavior of proteins within their natural cellular and tissue environments. This groundbreaking model was detailed in a paper published in Nature Methods. Unlike existing models that examine proteins in isolation, PINNACLE considers the broader context of tissues and cells where these proteins operate.
PINNACLE's approach to protein understanding
Marinka Zitnik, the study's senior author and assistant professor of biomedical informatics at the Blavatnik Institute at HMS, emphasized that "the natural world is interconnected, and PINNACLE helps identify these linkages." She further explained that this new tool could enhance our understanding of proteins and lead to safer medications. Unlike traditional models that provide context-free representations of proteins, PINNACLE recognizes that protein behavior can vary by cell and tissue type.
Potential in drug discovery
The researchers believe that PINNACLE could revolutionize our understanding of proteins' role in health and disease, and highlight new drug targets for designing more precise therapies. The tool is capable of finding which proteins engage in certain conversations and which ones remain silent within a specific cell type. This ability allows it to predict narrowly tailored drug targets for malfunctioning proteins that cause disease, potentially optimizing the drug discovery process.
PINNACLE's extensive protein representations
Using human cell data from a comprehensive multiorgan atlas, the scientists trained PINNACLE to produce graphic protein representations that encompass 156 cell types and 62 tissues and organs. To date, it has generated nearly 395,000 multidimensional representations. Each of the 156 cell types includes context-rich protein interaction networks of around 2,500 proteins. The current numbers of tissues, cell types, and organs are not the upper limits of the model.
