Ep199: Martin Burke on Making Small Molecule Medicines for the AI Era

Episode Summary

Executive Summary: Martin Burke traces his path from rural Maryland to pioneering block chemistry at Illinois, showing how modular, automatable synthesis can democratize medicinal chemistry. He argues small molecules remain uniquely powerful for drug discovery, and that AI plus robotics can compress discovery cycles from months to weeks while enabling new “molecular prosthetics” and better medicines such as antifungals and iron mobilizers.

Main Topics: Personal origin story and early scientific inspiration (Priority: 4/5): Burke grew up in rural Manchester, Maryland, in a family that strongly valued education and opportunity. A high school chemistry teacher and a family physician helped spark his interest in medicine and chemistry. From pre-med to chemical biology (Priority: 5/5): At Johns Hopkins and Harvard, Burke moved from aspiring physician to scientist as he saw chemistry’s creative power and its ability to solve real medical problems, especially through translational chemical biology. Molecular prosthetics as a therapeutic strategy (Priority: 5/5): Burke describes the idea of using small molecules to replace missing protein function, especially in cystic fibrosis, as a guiding concept for his work on amphotericin-derived compounds. Block chemistry and automation (Priority: 5/5): He explains block chemistry as a modular, repetitive carbon-carbon bond-forming approach that simplifies synthesis, enables robotics, and makes chemistry more accessible and scalable. Antifungal program and mechanism discovery (Priority: 5/5): Burke’s group overturned assumptions about amphotericin’s toxicity, showing the key mechanism was sterol binding and membrane sponge formation rather than channel formation, leading to safer antifungal candidates. Iron mobilization and other platform applications (Priority: 4/5): The lab’s platform also produced an iron-mobilizing small molecule discovered through a yeast screen, supporting work on anemia and opening broader applications for the chemistry platform. AI-enabled discovery and democratization of chemistry (Priority: 5/5): Burke argues that standardized blocks produce AI-friendly data, allowing active-learning loops with robots and models to build foundation models for molecules and potentially democratize chemistry beyond specialists.

Key Arguments: Small molecules remain a dominant and uniquely versatile medicine class because they are scalable, orally available, stable, and often inspired by natural products. Chemistry should move from artisanal, specialist-driven synthesis to modular, automatable workflows that anyone can learn and use. Block chemistry enables rapid iteration, which is the real bottleneck in drug discovery; shortening design-make-test cycles matters more than simply making one molecule faster. AI becomes far more useful when paired with standardized molecular building blocks that generate high-quality, learnable data. The antifungal toxicity of amphotericin was not primarily due to ion channels, contrary to decades of literature, but to sterol binding and membrane extraction. Molecular prosthetics can extend small molecules into spaces traditionally dominated by protein replacement or gene-based approaches. University ecosystems that support both basic science and startup formation are essential for translating discoveries into medicines. Democratizing molecule making could unleash outsider creativity and broaden the innovation base across medicine and materials.

Data Points: Hometown population context: More cows than people; one traffic light - Burke describes growing up in Manchester, rural Maryland. Scholarship opportunity: Full scholarship to Johns Hopkins via his father’s employer program - The Hudson Trust program covered tuition if accepted to select Maryland schools. MD-PhD model: 2-4-2 structure - Burke explains Harvard’s physician-scientist training path. Years of field history: About 200 years - He frames synthetic organic chemistry as a relatively young field. Number of reactions on the synthetic chemistry menu: About 1,000 reactions - Used to illustrate the artisanal complexity of traditional synthesis. Literature base on amphotericin toxicity: About 5,000 papers - Burke notes he reviewed extensive literature suggesting toxicity came from channels. Company funding: $95 million - Excelsior Sciences raised this amount to build the automated AI-enabled platform. Published automation paper: 2015 - He cites the paper showing iterative carbon-carbon bond formation could be automated. First iterative carbon-carbon bond formation demo: 2007 - He references the earliest proof-of-concept for block chemistry. Discovery/publication on iron mobilization: Science paper in 2017 - Tony Grillo’s yeast-screen-based discovery of an iron-mobilizing natural product. Cycle time for lead optimization: 3 to 6 months - Traditional chemist-driven iteration cycle before automation. Cycle time with Excelsior platform: 1 to 2 weeks - Automated AI-robot closed loop for design-make-test-learning. Founding of Midasyn: Started with 2 employees and $35K each on credit cards - Burke describes his first spinout on campus. Geographic scale of student collaboration: 8,000 miles - Middle school students in Pune, India printed molecules to Urbana, Illinois. Venture/support event: Third Rock pitch - Burke’s platform story led to support that helped launch Revolution Medicines.

Pivotal Quotes: "If it doesn't break the laws of physics, it's just a matter of time." — Martin Burke: Explaining Stu Schreiber’s influence on his scientific ambition and belief in solvable problems. "Rather than try to build machines that can do chemistry, we realized we had discovered chemistry that machines can do." — Martin Burke: Describing the conceptual leap from block chemistry to automation. "We're on a mission to democratize molecular innovation." — Martin Burke: Summing up the long-term goal of making chemistry accessible to non-specialists.

Implications: The interview suggests the next drug-discovery leap may come from combining modular chemistry, robotics, and AI to compress timelines and widen participation. If successful, this could accelerate new medicines and expand chemical innovation beyond elite labs.