COSMOS continues to engage with the growing public health challenge of adverse drug events, particularly drug-drug interactions (DDIs) that arise when multiple medications are taken concurrently. Our recent publication in Scientific Reports, published by Nature Portfolio, introduces a new model, the Protein Sequence-Structure Similarity Network (PS3N), for detecting dangerous drug-drug interactions (DDIs). Traditionally, it has been hard to predict these risky drug combinations because standard clinical trials cannot test every possible mix of medicines, especially over long periods or across diverse groups of people. Existing tools only rely on surface-level data like chemical traits or patient reports to make predictions. This research, however, introduces a deeper perspective by focusing on the underlying biology of how medications work.

Recently published in the journal Scientific Reports by Nature Portfolio, the study introduces the Protein Sequence-Structure Similarity Network (PS3N) to fix this problem. This model is the first to directly integrate both the genetic blueprints (called protein sequences) and the biological structures (called 3D protein structures) of drug targets to predict potential drug risks. By analyzing these fundamental biological building blocks, the model captures subtle molecular mechanisms that traditional methods often overlook.

The impact of this approach is demonstrated through its remarkable accuracy and real-world discovery. In rigorous testing across multiple datasets, key findings reveal that the model achieved up to 98% precision and 95% accuracy and successfully identified 297 entirely new drug interactions that have never been reported in existing clinical literature. 

The study further highlighted unseen dangers among these newly discovered risks, identifying that a common acne cream could potentially interact with treatments for serious conditions like heart disease or glaucoma. It also flagged unexpected risks when mixing certain mental health medications with treatments to help people quit smoking. By bringing these hidden biological connections to light, PS3N provides a powerful and reliable way to ensure safe and effective treatment outcomes for patients. Click here to read the full article.