Engineering Biology with AI: RNA Drugs & Photosynthesis Enzymes (Atomic AI + GigaCrop) скачать в хорошем качестве

Engineering Biology with AI: RNA Drugs & Photosynthesis Enzymes (Atomic AI + GigaCrop)

Presented by Playground Global at SF Deep Tech Week 2025 (https://luma.com/dtw2025-playground) https://www.playground.vc/ Deep Tech Week is a decentralized conference hosting the worlds best entrepreneurs and investors bringing science fiction technologies to life https://www.deep-tech-week.com/ ~~~~~~~~~~ A fast tour of how AI-native biology is designing RNA therapeutics and supercharging photosynthesis enzymes to boost crop yields—plus what skills the next generation needs. Playground Global’s Bin Kim hosts an Engineered Biology panel with Raphael (founder/CEO, Atomic AI) and Chris (founder/CEO, GigaCrop). Atomic AI builds foundational datasets and models to predict RNA structure & function, enabling rational design of next-gen RNA therapeutics—moving beyond trial-and-error screening. GigaCrop engineers enzymes to speed the CO₂→sugar steps in photosynthesis, lifting crop yields and drought tolerance; AI/ML guides residue selection, homolog choice, and screening to cut R&D time and cost. They dig into why data is the moat, how AI lets biology become an engineering discipline, specialized compute needs (atoms as the “data type”), and why the biggest impact may be quietly making old workflows cheaper and faster, not just flashy moonshots. Key takeaways Data is everything. Purpose-built, massive biological datasets are the real differentiator for AI models. Rational design arrives. Structure prediction (RNA/protein) enables targeted therapeutics and novel enzymes, not just “throw-and-see” screens. AI speeds & de-risks R&D. From residue/homolog choice to screen triage, ML cuts cost and timeline—unlocking projects that were uneconomic before. Novel function is now possible. AI can help design enzyme motifs not seen in nature, opening new reaction space. Bio becomes engineering. Scale, data, design, and reproducible pipelines—plus specialized compute stacks—define the next decade. Talent model. Go deep in one domain (bio or AI), speak the language of the other to collaborate effectively.