Genetics PhD @Stanford w/ @BrianHie | Prev @UMNComputerSci
Joined November 2018
- Tweets125
- Following407
- Followers269
- Likes168
Pinned Tweet
It was super fun working on Evo 2, a DNA language model trained on genomes across the tree of life!
Check out the preprint: arcinstitute.org/manuscripts…
A small 🧵 highlighting some mechanistic interpretability work on Evo 2 (Fig. 4) we did in collaboration with @GoodfireAI 🔥🔥🔥
Daniel Chang 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 🧵
Daniel Chang 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
New work from @_David_Li and @garykbrixi on large-scale co-evolutionary discovery across prokaryotic genomes! David and Garyk are incredibly talented + creative scientists and engineers, it was inspiring to watch their brainchild come to life
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.
Daniel Chang 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.
Daniel Chang 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.
Daniel Chang retweeted
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.
Daniel Chang retweeted
I am honored to have been selected as one of MIT Technology Review's Innovators Under 35 this year.
🔗ter.li/05p3ttfg
🔗technologyreview.com/superto…
This recognition is shared with the many amazing people I get to work with everyday as a @bioe_stanford PhD student and at @arcinstitute. Doing research alongside my brilliant labmates and advisor @BrianHie has been the best part of my PhD, and the science is truly a product of that.
My research has been driven by the insight that biology operates on design principles that can be leveraged to transform medicine, sustainability, materials, and beyond, and that computational modeling, including AI, will help us decipher those design principles at scale. Nature's innovations are written in genomes, and we have a long way to go in understanding them. You can read more about some of our efforts developing AI to understand and design genomes in my comment below.
Thank you to everyone at the @techreview and all the judges for organizing this, especially @AmyNordrum and @antonioregalado, for your support. I can't wait to see all the incredible things the other Innovators do!
Samuel is a force of nature and a huge inspiration!
Today in @ScienceMagazine, we report the first functional AI-generated genomes. 🧵
science.org/doi/10.1126/scie…
Daniel Chang retweeted
Thrilled to post thread re: new single-cell lineage of mouse embryo reconstructed w/ DNA Typewriter. One animal, zygote to late organogenesis (E13.5). Tree has 1,340,794 transcriptionally profiled, annotated tips (cells), 1,142,588 dated internal nodes, rooted at zygote
1/n
Daniel Chang retweeted
Excited to share our now published manuscript @CellCellPress highlighting advances in modeling the evolution of protein-protein interactions with MSA Pairformer. Big thanks to @ZhidianZ, @oliviatang116, @eunbelivable, Milot Mirdita, Martin Steinegger, and @sokrypton !🧵 (1/10)
cell.com/cell/fulltext/S0092…
Daniel Chang retweeted
Does every enhancer work with every promoter?
With @jengreitz and @WJGreenleaf, we revisit this long-debated question and resolve an outstanding contradiction in the field.
A tour 🧵👇
biorxiv.org/content/10.64898…
Daniel Chang retweeted
In new work, we lay out a vision for a high-level programming language for generative biology, called Proto.
Proto composes generative and predictive models spanning DNA, RNA, proteins, ligands, and their interactions, which we use to design complex biological functions. 1/n
Daniel Chang retweeted
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.
Daniel Chang retweeted
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…
Daniel Chang retweeted
Our lab is proud to present our latest work harnessing Bridge Recombinase for genome-scale editing in diverse bacteria, microbiome editing, and programmable horizontal gene transfer.
Bridge recombinase enables versatile rewriting of bacterial genomes biorxiv.org/content/10.64898… #biorxiv_synbio
Daniel Chang retweeted
Nature research paper: Transposable elements are driving rapid adaptation of Enterococcus faecium
go.nature.com/42qOQKc
Congrats @kenjmloi and @PeterHYoon on a very elegant story!
Daniel Chang retweeted
🧬Ultra fast petabase-scale virus discovery is here! 🚀
In our lab’s newest preprint, Jess explores the biodiversity of Papillomaviruses (PVs) in record time
🧵1/5 👇
🔗: doi.org/10.64898/2026
Daniel Chang retweeted
Submit to the GenBio Workshop at ICML this year!
Excited to introduce the 2026 workshop on Generative and Agentic AI for Biology at ICML 2026!
genbio-workshop.github.io/20…
1/5
Daniel Chang retweeted
Myloasm, our long-read metagenome assembler, is now published! w/ Max Marin & @lh3lh3
Very rewarding after > a year of development and countless hours thinking about assembly. Thanks to beta testers, Li lab, and reviewers for helpful feedback.
Link: rdcu.be/famFj
High-resolution metagenome assembly for modern long reads with myloasm - @lh3lh3 go.nature.com/3PBEwvR
