Biohub AI Model: Accelerating Drug Discovery & Protein Design

Biohub has unveiled an AI world model for drug discovery, aiming to enhance protein design. This initiative, backed by $500 million in funding, comes from the nonprofit biomedical research organization co-founded by Mark Zuckerberg and Priscilla Chan. The new open-source AI suite covers 6.8 billion proteins. It includes three AI models for mapping, predicting, and designing proteins. The centerpiece is the ESM Atlas, which alone covers those 6.8 billion proteins. The suite also features ESMFold2 for structure prediction and protein design, and ESMC, a protein language model. What's interesting is that these models can design functional binders with therapeutic-level affinity, and these results have been validated through laboratory testing. All three models are open-source, making them accessible to researchers with computational resources. This release is part of Biohub’s Virtual Biology Initiative, which has a total financial commitment of $500 million. The bottom line is that these open-source tools could significantly accelerate drug discovery and protein science for the entire field.