
Staff Research Scientist at Google DeepMind. Former adjunct assistant prof at @NYU_Courant. PhD at @mldcmu. ML for Bio/Chem (Prev. NLP). All opinions my own.
New York
Joined August 2012
- Tweets120
- Following3.5K
- Followers3.3K
- Likes456
Pinned Tweet
Really excited about our new work (led by @ptshaw2) on ProtEx: A Retrieval Augmented Approach to Protein Function Prediction.
Gives SOTA results on several tasks (especially for rare classes and sequences far from training set) to help better characterize the protein universe!
Excited to share new work from @GoogleDeepMind: “ProtEx: A Retrieval-Augmented Approach for Protein Function Prediction”
biorxiv.org/content/10.1101/…
Ankur Parikh retweeted
My team at Google DeepMind is hiring a student researcher (NYC, 2027) to work on how we evaluate sign language AI. Deaf and hard of hearing researchers are especially encouraged to apply, as are fluent signers. Get in touch: forms.gle/Gwfmp1cKGQ8oN2sSA
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/
Ankur Parikh retweeted
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/
Ankur Parikh retweeted
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. 🧵👇
Ankur Parikh retweeted
Combine a chemical knowledge, generative modeling, and structure prediction into a pipeline that can design enzymes for new-to-nature chemistry.
@ZvxyWu @alexechu_ @kelly_ruijie @ZiqiLi0513 @rob_fergus @francesarnold @pushmeet @jueseph
Ankur Parikh retweeted
This has been a wonderful collaboration with @GoogleDeepMind to make new-to-nature #enzymes. Great work from the whole team.
Introducing AlphaProtein Novo, a pipeline for de novo enzyme design! In collaboration with @francesarnold's group, we developed AP Novo and used it to get SoTA "novel-scaffold" activity on 2 benchmark reactions and create enzymes to synthesize the pharmaceutical motif piperidine and degrade the environmental toxin DEHP.
Work led by @ZvxyWu @ZiqiLi0513 with teammates @alexechu_ @kelly_ruijie @schulluc @jacobjinkelly @rosaliags @davidla @solserpens @j_reisenbauer, @pengliu_group, Josh, Thomas, Amy, Tristan, Harshnira, and others. It was a joy and privilege to work with you all!
Preprints and code:
biorxiv.org/content/10.64898…
biorxiv.org/content/10.64898…
github.com/google-deepmind/a…
Ankur Parikh retweeted
Today, our team is sharing a preprint on our new system for generative de-novo enzyme design, AlphaProtein Novo.
At @GoogleDeepMind, we have long focused on the building blocks of life (proteins). We developed AlphaFold to predict structure, trained language models to understand function, and built AlphaProteo to design de-novo binders. But proteins play many important roles in nature, such as enzymes that catalyze chemical reactions.
Designing entirely new enzymes from scratch to catalyze custom reactions has long been a holy grail for scientists. Our team has shown how AlphaProtein Novo can produce new-to-nature enzymes with state-of-the-art activity on 2 benchmark reactions. We’ve also designed custom enzymes to synthesize the pharmaceutical motif piperidine, and others to degrade the environmental plasticizer toxin DEHP.
While there is much more work to do in this direction, we’re excited to have demonstrated the ability to design functional proteins that go beyond what is known in nature. Huge congratulations to @jueseph, @ZvxyWu, Josh Abramson, the @GoogleDeepMind protein design team, and our collaborators in the @francesarnold & Liu Group! 🎉
🔗 Preprints and code:
biorxiv.org/content/10.64898…
biorxiv.org/content/10.64898…
github.com/google-deepmind/a…
In collaboration with the @francesarnold and @pengliu_group research groups, we @GoogleDeepMind are proud to release AlphaProtein Novo, a pipeline for small-molecule enzyme design.
We developed AlphaProtein Novo on a few standard model reactions, and validated designs in our own wet lab. We then wanted to challenge this system against nascent chemistries, so we asked leading scientists in biocatalysis whether we could partner on the reactions they cared about.
Introducing AlphaProtein Novo, a pipeline for de novo enzyme design! In collaboration with @francesarnold's group, we developed AP Novo and used it to get SoTA "novel-scaffold" activity on 2 benchmark reactions and create enzymes to synthesize the pharmaceutical motif piperidine and degrade the environmental toxin DEHP.
Work led by @ZvxyWu @ZiqiLi0513 with teammates @alexechu_ @kelly_ruijie @schulluc @jacobjinkelly @rosaliags @davidla @solserpens @j_reisenbauer, @pengliu_group, Josh, Thomas, Amy, Tristan, Harshnira, and others. It was a joy and privilege to work with you all!
Preprints and code:
biorxiv.org/content/10.64898…
biorxiv.org/content/10.64898…
github.com/google-deepmind/a…
The protein design team at Deepmind is hiring! We are looking for someone to run design campaigns and contribute to model improvement. Should have expertise in design applications and using wetlab data feedback to improve methods. Please RT! DM me with Qs. google.com/about/careers/app…
Ankur Parikh retweeted
We made FLIP2, a protein fitness benchmark spanning seven new datasets, including enzymes, protein-protein interactions, and light-sensitive proteins, as well as splits that measure generalization relevant to real-world protein engineering campaigns.
Ankur Parikh retweeted
Super proud of what this team is doing! And I can’t wait to share more soon.
We're thrilled to announce SignGemma, our most capable model for translating sign language into spoken text. 🧏
This open model is coming to the Gemma model family later this year, opening up new possibilities for inclusive tech.
Share your feedback and interest in early testing → goo.gle/SignGemma
Ankur Parikh retweeted
🚨🔔Foundational graph search task as testbed: for some distribution, transformers can learn to search (100% acc). We interpreted their algo!! But as graph size ↑, transformers struggle. Scaling up # params does not help; CoT does not help.
1.5 years of learning in 10 pages!
Ankur Parikh retweeted
Excited to share a new paper: “ALTA: Compiler-Based Analysis of Transformers” (w/ @James_Cohan, @jacobeisenstein, @kentonctlee, @JonathanBerant, @toutanova)
arxiv.org/abs/2410.18077
Ankur Parikh retweeted
Most protein function predictors only make predictions for labels seen in training.
We used LLM embeddings of text describing protein function to train ProtNote, which can generalize to new functional labels described in free text.
Code is available for everybody to try!
Ankur Parikh retweeted
Excited to announce our new work! 🧬 Some highlights are:
- sequences likelihoods predict zero-shot fitness capabilities
- a new method to calculate pLM likelihood in O(1) instead of O(L) forward passes
- providing a causal between training data and outputs
- suggesting a new finetuning method to improve pLM capabilities
(1/9)
biorxiv.org/content/10.1101/…
Ankur Parikh retweeted
Self-rewarding LMs at #icml2024 !
Thru iterative DPO (w/ a small amount of seed data), LLM instruction following ↑ (AlpacaEval 2.0, human, MT-bench) & reward modeling ↑ (corr w human rankings). @jingxu_ml will be presenting in Vienna (Tues 7/23 11:30am); please stop by! (1/2)
🚨New paper!🚨
Self-Rewarding LMs
- LM itself provides its own rewards on own generations via LLM-as-a-Judge during Iterative DPO
- Reward modeling ability improves during training rather than staying fixed
...opens the door to superhuman feedback?
arxiv.org/abs/2401.10020
🧵(1/5)
Ankur Parikh retweeted
Check out our new @GoogleAI paper: we curate a mixture of 5M human judgments to train general-purpose foundational autoraters.
Strong LLM-as-judge scores on RewardBench (87.8%), and highest perf among baselines on LLMAggreFact + 6 other benchmarks!
📰 arxiv.org/abs/2407.10817
👇
🚨 New @GoogleDeepMind paper 🚨
We trained Foundational Large Autorater Models (FLAMe) on extensive human evaluations, achieving the best RewardBench perf. among generative models trained solely on permissive data, surpassing both GPT-4 & 4o.
📰: arxiv.org/abs/2407.10817
🧵:👇
































