Today in Biology — Sep 16

[SOFIA] Welcome back to The Dish. We've got a lot to dig into today, especially on the national funding front, which always signals where the big pushes in biology are heading. And right at the top, it’s all about AI.

[DANIEL] Hmm. AI for science is definitely the headline. The Department of Energy’s Genesis Mission is really expanding, and it’s a significant move.

[SOFIA] Right? We're seeing new awards going out, with big names like the University of Washington and Old Dominion University getting involved. What I find really interesting is the DOE's emphasis on *open* AI models for science. For engineers like me, that's huge. It means fewer black boxes and more opportunities to actually build on what others are creating, rather than just consuming proprietary tools. It’s about making the underlying models accessible, which could really accelerate tool development.

[DANIEL] I agree on the openness. That’s critical for reproducibility and for the broader scientific community to validate and build upon these models. But the real question is how robust these open models will be, and what the validation framework looks like. It’s easy to say "open," but ensuring that the models are well-documented, rigorously tested, and interpretable for scientific discovery—that’s where the rubber meets the road. Are we just getting code dumps, or truly usable, well-characterized models?

[SOFIA] Exactly, the devil's in the details. But still, the intent for open models is a positive signal. And speaking of AI, ARPA-H just launched this ambitious project to build the world's first FDA-authorized clinical AI specifically for cardiovascular care.

[DANIEL] That’s a very high bar. "FDA-authorized" means rigorous clinical trials, clear endpoints, and demonstrable safety and efficacy. Cardiovascular care is an area with immense clinical need, from early diagnosis of subtle heart conditions to optimizing treatment plans. An AI that could reliably interpret complex imaging, patient histories, and genetic data to flag risks or recommend interventions *before* a major event would be transformative. But getting that authorization means showing it performs better than current gold standards, without introducing new biases or errors, across diverse patient populations.

[SOFIA] Absolutely. That's not just a fancy algorithm; that's something that could directly impact patient outcomes. It’s moving AI from research into direct clinical application. And it’s not just AI. The NSF is also making some big moves. They’ve just funded a new center for understanding organismal resilience and adaptation, which is a $20 million award.

[DANIEL] Organismal resilience is fascinating, especially when we talk about engineering biology. Understanding how organisms adapt to stress—thermal, chemical, nutrient limitation—provides crucial insights into robust design principles for synthetic systems. This isn’t just about observing nature; it’s about extracting the underlying rules of robust biological function.

[SOFIA] Right, for someone trying to engineer a microbe to produce a new biofuel, knowing how to make it resilient to the industrial conditions it'll face is everything. This kind of research helps us understand the fundamental mechanisms. It’s like getting the instruction manual for stress tolerance.

[DANIEL] And on the funding landscape generally, there’s chatter about the NSF overhauling its funding approach to align with White House priorities. *Nature* reported on it, and there's some concern among staff about how new initiatives will be funded without clear indications of new money. That always raises questions about whether existing programs will be cannibalized.

[SOFIA] Yeah, that’s the flip side of big new initiatives, isn't it? Will the new priorities mean less support for fundamental, curiosity-driven research that doesn't immediately fit a national strategic goal? That's always a tension point. It’s great to see focused investment, but we also need that broad base of discovery.

[DANIEL] A tension that isn’t going away soon.

[SOFIA] Definitely not. Alright, that’s it for Today in Biology. Next up, we’re diving into…