BioE PhD student @stanford and @arcinstitute in the Hie Lab // AI for biological design
Joined April 2020
- Tweets70
- Following405
- Followers1.1K
- Likes226
Pinned Tweet
Biological engineering has long aimed to design living systems with the programmability of software, but the tools to do so rarely fit together.
Today, we’re sharing our preprint introducing Proto, a high-level programming language for generative biology that composes models and design tools across DNA, RNA, and proteins to optimize towards user-specified functional objectives.
Aditi Merchant retweeted
Language models can learn new tasks from just a few examples in their prompt, with no retraining. This is called in-context learning (ICL). Now genome language models can do it too.
With a handful of examples, Evo 2 ICL generates proteins, de novo antibodies, CRISPR-Cas9 loci, mitochondrial genomes and entire bacteriophages.
We can even use the synthetic sequences iteratively to construct a chain-of-thought (CoT) and achieve harder design goals.
New work with Yukun A. Hao, @BrianHie and team 🧵
Aditi Merchant retweeted
In new work, we report a surprising finding: prompting Evo 2 with a handful of sequences of shared function can steer the model to generate new sequences with similar function, a form of generative in-context learning (ICL). This is an emergent phenomenon that we did not anticipate during training.
In a first study, led by Alex Hao, we show that ICL enables few-shot design of mammalian regulatory elements, including novel enhancers with cell-type specificity for a neuronal subtype in the mouse brain.
In a second study, led by @garykbrixi and Alex Hao, we extend these results to the computational design of natural proteins, de novo antibodies, protein-RNA systems, and complete bacteriophage genomes.
As part of this work, we also release several useful new artifacts, including a highly efficient new inference stack. 1/n
Aditi Merchant retweeted
Launching Juno, your second brain in iMessage.
I collect little things all the time. A ramen spot from a TikTok. An article on TechCrunch. A chair I saw on Instagram. A screenshot of a friend telling me to read something. An idea at 1am.
Pretty soon it's scattered everywhere, like in Apple Notes or in texts to myself. Then I never find it again, let alone do anything with it.
So I built Juno.
Juno is a contact in your phone. Text it anything you wanna keep, a link, a photo, a screenshot, a voice memo, a half thought, and it files it into the right note for you. No app, no tags, no "where should this go".
When I need something I just ask, like "what was that chair I saved?" Everything's on the web too, already organized into notes for you.
After a while I realized Juno knew my taste well enough to actually get stuff done. So I let it. It orders things I saved, books the spots I've been meaning to try, texts me a brief every morning, and built me a site for a weekend plan. I've even started to call it.
It's proactive too. It sets up routines on its own, like checking in after plans or a heads up before a friend's birthday.
I've been texting Juno every day for a couple weeks and it's been really useful. If it sounds like something you'd use, try it, it's free.
Aditi Merchant retweeted
Today in @Nature, we introduce SynthID Bio, a family of watermarking methods for protein sequences and structures, that we believe will contribute to responsible development of AI for biology. Learn more in our blog and this thread 🧵 - deepmind.google/blog/introdu…
A truly beautiful paper by the always inspiring (and wise🦉) @_David_Li and @garykbrixi! Check out how you can leverage learned co-evolutionary relationships for biological discovery with Minerva!
Excited to share Minerva, our approach using genome language models for biological discovery! Using Minerva, we find that UG27 reverse transcriptase systems encode variable arrays of diverse ncRNAs with a shared structure, each templating a short DNA hairpin. With @garykbrixi.
Aditi Merchant retweeted
As with biological design, AI can also accelerate biological discovery.
In new work led by @_David_Li and @garykbrixi with @mfgrp, we show how genome language models enable systematic discovery of genetic elements, revealing a new class of reverse-transcriptase mechanisms.
Aditi Merchant retweeted
Genomes are full of dark matter of unknown functions.
We present Minerva, a new method for discovery guided by genome language models. Minerva reveals the interactions hidden in non-coding DNA, pointing to hundreds of new putative RNAs and repetitive elements per bacterial genome. Minerva allows us to find and study elements invisible to traditional methods at orders of magnitude greater scale than before.
You can now run Proto’s suite of 100+ Bio AI, database search, and bioinformatics tools directly from ChatGPT! Excited to see what everyone builds 🧬
More features and updates coming soon, stay tuned :)
Learn more about the Rosalind Workbench research preview below!
Rosalind Workbench connects scientific questions to specialized models, tools, and reviewable outputs in one workflow, from protein structure and sequence analysis to sequencing pipelines.
developers.openai.com/blog/r…
Aditi Merchant retweeted
You can now use Proto in ChatGPT!
Here is de novo nanobody design with Germinal (chatgpt.com/share/6a919bda-e…) and semantic design with Evo 2 (chatgpt.com/share/6a90a552-c…).
Check out the Rosalind Workbench Research Preview: developers.openai.com/blog/r…
Rosalind Workbench connects scientific questions to specialized models, tools, and reviewable outputs in one workflow, from protein structure and sequence analysis to sequencing pipelines.
developers.openai.com/blog/r…
Aditi Merchant retweeted
Excited to share the final version of ProteinDPO is out today in @naturemethods! New in the paper is application of the model to stabilization of the pre-fusion state of hemagglutinin, the primary component of flu vaccines. 1/8
🔗nature.com/articles/s41592-0…
📑rdcu.be/fzUc2
Aditi Merchant retweeted
Our work led by @talaldotpdb on aligning protein generative models to experimental fitness with ProteinDPO is now published in @naturemethods!
New in the published paper includes stabilizing mutations to H5 hemagglutinin, a component of flu vaccines.
nature.com/articles/s41592-0…
Aditi Merchant retweeted
Following the publication of our phage gen paper, I wrote some personal thoughts on biosafety and biosecurity as generative biology improves.
Given the trajectory of progress, there should be concern and urgency -- but also optimism.
brianhie.com/present_and_fut…
Aditi Merchant retweeted
At Therna we treat RNA as a language you can read, score, and rewrite.
Our new post: designing a glucagon mRNA for HepG2, with Evo 2, PARADE, EnCodon and ViennaRNA in one loop using Proto.
Illustrative only. Public models, no wet lab validation.
Link in the comments
Happening today! Tune in to learn more about Proto and how to use it for your own biological design goals 🧬
We’re excited to host @aditimerch on Tuesday, July 21, for a presentation on her recent work, “A High-Level Programming Language for Generative Biology with Proto.” The talk will take place from 6:00–7:00 p.m. ET / 3:00–4:00 p.m. PT
Aditi Merchant retweeted
Check out our website proto.evodesign.org, which visualizes our designs and provides open access to 120+ biological AI models and tools!
To learn how to navigate the UI and start designing with our Proto agent, check out youtu.be/8UnOrC_0jmo?si=rq-Q…
We've received amazing feedback over the past day and we're excited to see what the community builds! proto.evodesign.org
Huge congrats to the Germinal team on their beautiful paper! An exciting day in AI for biological design!
Germinal is out today in Nature Biotechnology! Important additions to VHH design from the preprint: we demonstrate scFv design, evaluate binder polyreactivity, and characterise epitope specificity by alanine mutagenesis and a beautiful cryo-EM structure… nature.com/articles/s41587-0…
Aditi Merchant retweeted
Imagine programming protein, DNA, and RNA systems like you would write computer code, or even by natural language prompting of an AI agent. @brianhie and team just made this a reality with Proto: a high-level programming language for generative biology.
And check out Daniel’s post to learn a bit more about the language abstraction Proto introduces and how to run Proto for your design campaigns from your browser!
Biological engineering has long aimed to make biological systems designable, composable, and programmable across scales and modalities. However, existing biological AI models and tools are siloed and difficult to compose.
Today, we’re proud to share our preprint for Proto, a high-level programming language and open infrastructure layer generative biology.
*This thread is centered around the Proto Language. For the full paper thread, check out: nitter.cf/aditimerch/status/2069…
