Computational bio at @ProfluentBio | @UCSF PhD @JoeBondyDenomy lab
San Francisco
Joined December 2017
- Tweets551
- Following1.2K
- Followers376
- Likes10.9K
Pinned Tweet
Excited to share an exciting discovery which I think will help us understand the anti-phage arsenal of bacteria!
biorxiv.org/content/10.1101/…
1/
Matt retweeted
I'm excited to share that our work on co-translational degradation of cas12a mRNA is finally out! Many thanks to all the authors who made it happen.
nature.com/articles/s41586-0…
Matt retweeted
Protein language models just got an upgrade. Meet Profluent-E1: a free, flexible, frontier protein sequence encoder.
E1 is built with retrieval augmentation to learn from multiple sequences. Models trained over 4T tokens with only 150M-600M params, E1 is SOTA for zero-shot functional and unsupervised structural tasks. It raises the bar for protein representation learning and is freely available today.
Matt retweeted
Excited to share our last collab paper, now out in print @ScienceMagazine!🎉
We reveal how dITP synthesis activates a SIR2 NADase effector to block phage infection! 🛡️🧬
science.org/doi/10.1126/scie…
Matt retweeted
Check out our new paper describing rAcrVIA1: an RNA anti-CRISPR that mimics guide RNAs to inhibit Cas13. Very fun project with stellar UW undergrad Victoria Hayes and collaboration with Ning Jia's group! science.org/doi/10.1126/scie…
Matt retweeted
What could scaling unlock for biology?
Introducing ProGen3- our next AI foundation models for protein generation. We develop compute-optimal scaling laws up to 46B parameters on 1.5T tokens with real evidence in the wet lab.
+we solve a new set of challenges for drug discovery
Matt retweeted
Claire's paper now entitled "Multi-interface licensing of protein import into a phage nucleus" is now published here: nature.com/articles/s41586-0…
@CKokontis used classic genetics to figure out how proteins are selected for import into a "phage nucleus"
Some notable additions:
New preprint from the lab, led by outstanding student Claire Kokontis! (@CKokontis) In this work Claire et al. discover Imp1-Imp6, phage proteins required for the selective import of proteins into the jumbo phage nucleus! Imp1 (below) is the special one.
biorxiv.org/content/10.1101/…
Matt retweeted
Can LLMs exert molecular control in modulating protein-DNA interactions?
New research where we demonstrate AI can reprogram DNA recognition of gene editors-- all in single-shot without iterative laboratory screening.
Incredible positive impact to unlock
Matt retweeted
🌟Our latest results are now published at NAR🌟 tinyurl.com/bdfbdxt4. We hope that AntiDefenseFinder will enable scientists to further uncover the natural diversity of inhibitors (or regulators) of known bacterial defense systems. Recap below!
Excited to introduce AntiDefenseFinder - a free, open source tool that detects anti-defense systems in any genomic sequence. Available here to toggle on/off: defensefinder.mdmlab.fr/. Read below for our insights into the >47K detectable systems📈 tiny.cc/8xrizz
Matt retweeted
I am very proud to share our paper on Tad1/Tad2 sponges, which appeared to be multifunctional anti-defense proteins that inhibit many different signaling-based immunity systems!💫
nature.com/articles/s41586-0…
Matt retweeted
I'm thrilled to share that I've joined Dr. Pam Ronald’s lab at @ucdavis to study how plants evolve multi-pathogen resistance, and in the long-term, build upon sustainable agriculture practices. As pivot into this field, I'd love to connect w/ folks in the ag & climate space!🌱🌎
Matt retweeted
Tomorrow, 9/17 @ 4pm ET, we'll have @jeffruffolo @ProfluentBio present ProseLM: Adapting protein language models for structure-conditioned design
Sign up on our website to receive zoom links via email or DM us to join!
Read the paper: biorxiv.org/content/10.1101/…
Matt retweeted
The final version of our paper on anti-CRISPRs "in disguise" as Cas proteins is out today in Nature! Congrats to Mark, Edith and the whole team! nature.com/articles/s41586-0…
Matt retweeted
My favorite and hardest NMR experiment ever conducted is finally out! Congrats to everyone involved, but especially Daphne for pushing the project to the end 🎉🎊
Anti-CRISPR enzyme is widespread, specific and potent: as good or better for lytic rep and lysogen stability than a strong non-enzyme! Long running effort in @fraser_lab/BD lab with many contributors (co-led by Daphne Chen and @leahtangroe).
biorxiv.org/content/10.1101/…
Glad this is finally out! 🛡️AntiDefenseFinder 🔎🦠 will identify over 150 defense inhibitor families across MGEs, phages, and prokaryotes. Update DefenseFinder to v1.3.0 to use. Congrats Florian and @ErinHuiting!
Excited to introduce AntiDefenseFinder - a free, open source tool that detects anti-defense systems in any genomic sequence. Available here to toggle on/off: defensefinder.mdmlab.fr/. Read below for our insights into the >47K detectable systems📈 tiny.cc/8xrizz
Matt retweeted
For antibodies, we started with the clinically approved PD-1 binder nivolumab, focusing design on either the CDR loops or framework. We obtained a wide range of binders from each approach, including two antibodies with nearly 3-fold higher affinity for PD-1 than nivolumab.
Matt retweeted
For gene editors, we started with a low-activity base editor previously designed using LMs, then focused design on either active site or non-active site residues. Both yielded base editors with 50% higher A-to-G editing, approaching SOTA editors in just one round of optimization.
Matt retweeted
Protein structural search uncovers an ancestral Cas13 protein led by the brilliant @PeterHYoon !!
Check out the paper and thread! 👇
Excited to share our work published in @ScienceMagazine today answering the question: “Where does Cas13 come from?”
science.org/doi/10.1126/scie…
