Growing GPUs on trees: design of functional protein–semiconductor hybrid materials
summary
Computational pipeline uses RF diffusion to generate protein backbones, then sequence models and millions of in silico filters produce a library—35,000 designs—that is expressed in yeast and tested for templating semiconductors. Specific successes include designed zinc oxide cages that trigger ZnO nucleation (fluorescence kinetics, XRD confirmation) and cryo‑EM showing ordered lattice within protein assemblies. Work extends to silica masks for UV lithography with claimed sub‑nanometer potential and ongoing efforts to scale via engineered microbes.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 2:15Nature’s blueprint
- 6:36Protein design pipeline
- 10:11Proteins template semiconductors
- 16:35Functional architectures
- 19:51Controlled material growth
- 24:31Helical structures and electronics
- 27:45Genetically encoded lithography masks
- 30:36Bio-mining vision
- 33:30Q&A
related talks
48mReprogramming Biology
46mMachine Learning-based Design of Proteins and Small Molecules
39mDesigning 1000 genomes in a single bound: The Codon DSL
18mGenerative Design
53mHow machine learning is influencing protein engineering
10mGenesis Final Round: Nanyang Biologics
29mBiomaterials as UI
32mMachine Learning-Based Design Of Proteins