


Research Scientist at Google DeepMind


Watch how @GoogleDeepMind chemists used Antigravity to extract compound data from patents and papers, analyze what drives potency, and propose new candidate molecules with synthesis routes in just an hour. The agent also launches an interactive viewer so chemists can explore the data in a collaborative way: browsing retrieved molecules, inspecting their properties, and plotting how molecular structure shapes potency.

Tools for Gemini/Antigravity to do chemical reasoning: retrosynthesis, reaction mechanism proposal, etc. Great for enzyme design / engineering and metabolic engineering!
Have a fun chemistry challenge? Find a talented collaborator in Google Antigravity 🚀 Today we're introducing Cheminformatics Skills 📖 Blog:lnkd.in/eGX-q8jE 🧪 Try it: Install the Science plugin in Antigravity: lnkd.in/etVfMGEb 🧵 1/

Have a fun chemistry challenge? Find a talented collaborator in Google Antigravity 🚀 Today we're introducing Cheminformatics Skills 📖 Blog:lnkd.in/eGX-q8jE 🧪 Try it: Install the Science plugin in Antigravity: lnkd.in/etVfMGEb 🧵 1/

At @GoogleDeepMind, we believe one of AI's most transformative impacts is empowering scientists to solve society's hardest problems. Over the past year, we've evolved our coding agent, @Google @Antigravity, to accelerate end-to-end scientific workflows combining our latest Gemini models with science-specific: • breakthrough models like AlphaFold and AlphaGenome • new interfaces like the Biomolecular Structure Viewer The impact of this integration has been transformative for the pace of our research. As an example, in a single session using Antigravity, our researchers working on antimicrobial resistance surfaced a novel antimicrobial peptide candidate. To support academic researchers leverage these tools, we've launched an Antigravity for Science credits program and are looking forward to what this enables: antigravity.google/use-cases… Read the antimicrobial resistance case study: deepmind.google/blog/acceler…

Really excited to see where we can take this next, stay tuned for more! Read more about Gemini Argon in the blog post: goo.gle/3W85sXj

It is surreal to see this workflow compressed to a few sessions of prompting with Argon. This was joint work with @ank_parikh, @kelly_ruijie, @jueseph, @alexechu_, Michael, Josh, Luca, Amy, Tristan, Harshnira, Lucy, and the rest of the protein scientists at GDM.

These went through an initial round of wet-lab testing and we found the desired catalytic activity across these designs.

Argon assembled training data, fine-tuned a multi-family protein language model and created sequence & geometric based active-site filters to select from the outputs of that model. The final result was 5 divergent, novel candidates (59-81% sequence identity to natural enzymes).

Enzymes like these belong to a class called serine esterases, which are used to manufacture pharmaceuticals or biofuels, formulate detergents and even recycle plastics.

We worked with Argon to design enzymes to cut a particular ester bond in 4MU-acetate, a reaction that allows for a simple fluorescence assay.

Designing functional enzymes beyond natural sequence space can involve months of literature mining, data curation, bespoke model tuning, and candidate filtering.

Over the past few weeks, we have been putting Gemini 4 Argon to the test on protein engineering workflows at Google DeepMind, and we have some interesting initial (zero shot!) results to share. 🧵👇