Today in Biology — Aug 19

[THEO] Alright, good morning, Cultivarium listeners! Theo here, kicking off 'Today in Biology' with what's bubbling up in the big picture. And if you've been feeling like the ground beneath your grant applications is shifting, you're not wrong.

[DR. MARA] Indeed, Theo. We're seeing a significant re-evaluation of how fundamental research is being conceptualized and funded, particularly within the US federal agencies. It’s not just about what science gets done, but how it’s enabled.

[THEO] Exactly! And the biggest splash right now has to be the Department of Energy, the DOE, really leaning into AI for science with their new Genesis Mission. They're looking for contributors to build open-weight AI models. Now, for those of us who don't spend our days wrestling with supercomputers, "open-weight foundation models" means they're not just building an AI and keeping it secret; they're making the core components, the "weights," available for anyone to build on. It’s like giving everyone the blueprint and the basic parts for a really powerful new engine, instead of just selling the finished car.

[DR. MARA] Precisely. The goal here is to accelerate scientific discovery, especially where large-scale data analysis and complex simulations are critical. Think about materials science, climate modeling, or even systems biology. These fields generate immense datasets that traditional analytical methods struggle to fully exploit. An open-weight model means researchers across various institutions can adapt and fine-tune these powerful AI tools for their specific, often unique, scientific problems, rather than having to build from scratch.

[THEO] Right. And this isn't just about making existing algorithms faster. The idea is that these AI models, once trained on vast scientific datasets, might uncover patterns or relationships that human scientists, no matter how brilliant, simply can’t perceive. It’s an augmentation of scientific intuition, not a replacement. And for engineering biology, that could mean things like predicting protein folding or designing novel enzymes with unprecedented accuracy.

[DR. MARA] It holds that potential, yes. The emphasis on "scientific datasets" for pre-training and fine-tuning is crucial here. Unlike general-purpose AI, these models are intended to embed scientific principles and constraints directly into their architecture, making their predictions more physically and biologically plausible.

[THEO] Okay, so big picture: more open-source, powerful AI coming from the DOE. But let's pivot to another huge player: the National Institutes of Health, the NIH. They're proposing a major revamp of how they score grant proposals. This is a pretty big deal for anyone trying to get their research funded.

[DR. MARA] It is. The NIH's grant review process is the gatekeeper for an enormous amount of biomedical research in the US. How they score proposals directly dictates what research gets funded and, by extension, what scientific directions are prioritized. The proposed changes suggest an acknowledgment that the current system may not be optimally serving the scientific community or public health needs.

[THEO] And we're seeing some real friction there. I've seen a lot of chatter online about grant funding being held up, with some calling it a "graveyard for grants." There's real concern that even excellent proposals, including for things like maternal health, are getting stuck. It raises questions about the efficiency and responsiveness of the current system, particularly when we have agencies like the NSF simultaneously announcing new funding opportunities — over a billion and a half dollars for foundational research, actually.

[DR. MARA] That's a critical point, Theo. The juxtaposition of significant new funding announcements from the NSF, across engineering and mathematical and physical sciences, with reports of NIH grant bottlenecks, highlights a potential disconnect in the overall funding ecosystem. It indicates where the national scientific priorities might be shifting or where existing structures are struggling to adapt to the pace of scientific need. For engineering biology, this means paying close attention to which agencies are actively seeking proposals and whether their review processes are agile enough to move novel, interdisciplinary work forward.

[THEO] So, in short: the DOE is pushing hard on open AI for science, the NIH is wrestling with how it scores and processes grants, and the NSF is out there with a hefty chunk of new foundational research money. It's a dynamic landscape, and researchers need to be strategically aware of these currents.

[DR. MARA] Absolutely. Understanding these shifts is key to navigating the research funding environment effectively.

[THEO] Alright, on that note, we’ll take a quick break. When we come back…