For years, computational biologists have relied on sophisticated scoring functions to predict how proteins in the human body bind together — a process fundamental to fighting disease, transporting molecules, and keeping our cells functioning. But a team at Yale University has discovered something surprising: when it comes to predicting protein binding, less may actually be more. Their simple model, using just two physical features, outperforms seven of the most advanced methods currently used in the field. The findings, published in Physical Review E, could accelerate the hunt for new medicines and deepen our understanding of the roughly 10,000 distinct proteins working inside us at any given moment. When two proteins bind to form what's called a heterodimer, they fit together like pieces of a biological puzzle — but not every piece will match. Predicting which combinations will actually connect, and where, is the central challenge in the field. The stakes are high: better predictions could reveal how proteins contribute to disease and point researchers toward entirely new therapeutic targets. Currently, high-resolution structural data exists for only a couple thousand of the body's proteins, leaving a vast landscape unexplored. "We have tons of proteins in our bodies, but we don't know what complexes they form, what cell functions they carry out or if they are implicated in disease," said Professor Corey O'Hern, who led the study. Lab experiments to test protein pairs one by one are prohibitively slow and expensive, making computational models essential. Yet O'Hern and his team found that existing scoring functions, despite appearing accurate in earlier assessments, fell short under rigorous testing. "When the previous literature assesses current methods for scoring computational models, they claim that they're accurate, but when we do rigorous tests, we find that they're not," he said. The Yale team's solution is striking in its simplicity. Their support-vector regression model considers just two features: the size of the interface where the two proteins would connect, and how well intertwined the proteins are at that junction. This minimal approach matched or exceeded the performance of the seven leading scoring functions. "In the current study, we focused on the simple problem of identifying the binding interface between two rigid protein pairs," O'Hern said. "In the future, we will develop methods for identifying binding pairs when we do not know the bound form of the monomers." The work represents more than an academic curiosity — it's a potential turning point in the slow, expensive process of understanding protein function. As researchers build better tools to map the body's molecular puzzle, the path from basic science to new treatments grows a little clearer.