{"id":69997,"title":"🧪 ⚛️ Diffuse Bio: Generative AI for protein design","tagline":"At Diffuse, we build new therapeutics, vaccines, and enzymes with AI!","body":"Hi 👋🏾 I’m Namrata, founder of ⚛️⚛️⚛️ **Diffuse Bio!** ⚛️⚛️⚛️\n\nAt Diffuse, we computationally design therapeutics, vaccines, and enzymes better, faster, and cheaper — but most importantly, we generate **molecules that simply can’t be designed with existing methods today**.\n\n### The Problem:\n\nProteins are macromolecules that mediate a significant fraction of the cellular processes that underlie life. An important task in bioengineering is designing proteins with specific 3D structures and chemical properties which enable targeted functions. \n\nSo far, computational protein design methods have had some success — but have enormous limitations, the primary one being **low success in downstream validation experiments**. These methods are also painfully slow and scale poorly.\n\n### What we do:\n\nFor the past several years, I’ve been working on **re-envisioning the computational protein design toolkit with AI** in my PhD and beyond, with the goal of overcoming these challenges. Some highlights are the first crystal structures of [AI-designed proteins](https://www.nature.com/articles/s41467-022-28313-9) (i.e. experimental validation) and the **first diffusion models** for [protein structure and sequence generation](https://nanand2.github.io/proteins). Below you can see one of our early models producing structures unconditionally from noise!\n\n![uploaded image](/media/?type=post\u0026id=69997\u0026key=user_uploads/459396/08cd90de-6450-4ebe-8873-a77d625e0ed1)\n\nNow, we’re **scaling up these methods** and **applying them to grand challenge problems in molecular design**. We’re able to handle a whole array of protein engineering tasks at the _push of a button_ — from loop redesign and sequence engineering, all the way to binder design and _de novo_ structure generation. \n\n### Curious to learn more?\n\nCheck out our coverage in the [**New York Times**](https://www.nytimes.com/2023/01/09/science/artificial-intelligence-proteins.html#:\\~:text=Namrata%20Anand%2C%20a%20former%20Stanford%20University%20researcher.%20She%20is%20now%20building%20a%20company%20in%20generative%20A.I.%20protein%20design) and [**NBC news**](http://bit.ly/3JjgVKL), and get in touch at [info@diffuse.bio](mailto:info@diffuse.bio)!\n\n### We’d love your help!\n\n* If you’re working on any protein or molecular engineering problem, we’d love to hear from you!\n* We’d love **intros to pharma and biotech companies** working on problems in protein therapeutics, antigen design, and enzyme design. \n* We’d also love intros to any companies (preferably early stage) doing **high-throughput protein characterization** **assays** (yeast display, phage display, etc).\n* Sign up [here](https://docs.google.com/forms/d/e/1FAIpQLSdamv0S3LB2cyVLJXT6QdCG2g71Lobwkw59qTYGullYgqg7bw/viewform?usp=sf_link) to **beta test** our first-gen protein design software!","slug":"ICz-diffuse-bio-generative-ai-for-protein-design","created_at":"2023-03-07T05:47:09.229Z","updated_at":"2026-05-25T00:30:34.521Z","total_vote_count":64,"url":"https://www.ycombinator.com/launches/ICz-diffuse-bio-generative-ai-for-protein-design","share_image_url":"//bookface-static.ycombinator.com/assets/ycdc/yc-og-image-c440a0ad1dacfb86eeeb343717479cc54d256614449b4ef719977a0a451f8bc8.png","company":{"id":28037,"name":"Diffuse Bio","slug":"diffuse-bio","url":"http://diffuse.bio","logo":"https://bookface-images.s3.amazonaws.com/small_logos/961a0aa4b25ffd0049a0526af8ad2406763125f4.png","batch":"Winter 2023","industry":"Healthcare","tags":["AI-powered Drug Discovery","Deep Learning","Generative AI","Machine Learning","Biotech"],"search_path":"https://bookface.ycombinator.com/company/28037"}}