Professor James Collins, a pioneer in synthetic biology at MIT, is redefining drug discovery by merging computational power with experimental biology. By leveraging deep learning, his team identifies potent antibiotics like halicin, which can neutralize multidrug-resistant bacteria. This interdisciplinary approach combines network biology and systems microbiology to tackle global health crises more proactively than traditional methods allowed.

In a landmark study, the Collins Lab utilized generative AI to architect entirely new molecules from scratch. By employing variational autoencoders—AI models that learn to create new data similar to their training set—the team generated millions of candidates. This process involved filtering molecules based on chemical feasibility and selective activity, eventually producing lead compounds like NG1, which specifically targets resistant gonorrhea while sparing beneficial bacteria.

To bridge the "valley of death" between laboratory discovery and clinical use, Collins co-founded Phare Bio. This nonprofit works alongside the Antibiotics-AI Project and receives support from ARPA-H to develop pre-clinical candidates. By integrating generative models with high-throughput biological testing (automated systems that screen thousands of compounds simultaneously), the initiative aims to build a rapid-response pipeline for global health threats.

Professor James Collins, a pioneer in synthetic biology at MIT, is redefining drug discovery by merging computational power with experimental biology. By leveraging deep learning, his team identifies potent antibiotics like halicin, which can neutralize multidrug-resistant bacteria. This interdisciplinary approach combines network biology and systems microbiology to tackle global health crises more proactively than traditional methods allowed.

In a landmark study, the Collins Lab utilized generative AI to architect entirely new molecules from scratch. By employing variational autoencoders—AI models that learn to create new data similar to their training set—the team generated millions of candidates. This process involved filtering molecules based on chemical feasibility and selective activity, eventually producing lead compounds like NG1, which specifically targets resistant gonorrhea while sparing beneficial bacteria.

To bridge the "valley of death" between laboratory discovery and clinical use, Collins co-founded Phare Bio. This nonprofit works alongside the Antibiotics-AI Project and receives support from ARPA-H to develop pre-clinical candidates. By integrating generative models with high-throughput biological testing (automated systems that screen thousands of compounds simultaneously), the initiative aims to build a rapid-response pipeline for global health threats.