Kevin Wu
@Kevin_E_Wu
Followers
340
Following
67
Media
0
Statuses
34
AI Research Scientist at @chaidiscovery | Formerly at @stanford, @ucberkeley. All views are my own. 🇹🇼
San Francisco
Joined November 2020
after months of antibody design papers that only work on single chains, we are seeing much-needed progress on full IgG congrats to the chai team!
Today, we’re releasing new data showing that Chai-2 can design antibodies against challenging targets with atomic precision. >86% of our designs possess industry-standard drug-quality properties without any optimization. Thread👇
2
15
175
Today, we’re releasing new data showing that Chai-2 can design antibodies against challenging targets with atomic precision. >86% of our designs possess industry-standard drug-quality properties without any optimization. Thread👇
19
116
470
🧬 Dr. Mikael Dolsten is joining the @ChaiDiscovery board! In 15 years as @Pfizer's Chief Scientific Officer he: ⚛️ Progressed 150 molecules to clinical trials 💊 Delivered 36 approved medicines, from vaccines to cancer therapies
4
12
108
Some use cases are cooler than others, and it's safe to say zero-shot antibody design is one of them. Congrats on launching Chai-2!🧬
Big thank you to stellar engineering team at @modal_labs, who provided some critical infrastructure for this effort. They are a joy to work with.
0
6
43
Truly a privilege to have worked with everyone else on the team to bring this vision to reality! Although we’ve made a ton of progress there’s still so much left to do. On to the next challenge!
We’re excited to introduce Chai-2, a major breakthrough in molecular design. Chai-2 enables zero-shot antibody discovery in a 24-well plate, exceeding previous SOTA by >100x. Thread👇
0
1
12
Chai-1 has always been available for commercial use via our server. Today, we're also making Chai-1(r) code and weights available under an Apache 2.0 license, which permits broad commercial use. https://t.co/EFz6edbreN
github.com
Chai-1, SOTA model for biomolecular structure prediction - chaidiscovery/chai-lab
5
61
276
What a week! First, @GoogleDeepMind fulfilled its #AlphaFold3 promise, and now @chaidiscovery delivered a long-awaited killer feature - custom distance restraints: https://t.co/LEPpdrM3fG! Here is a 7SYZ (AB-protein) example (x-ray in white).
3
6
41
Today we’re releasing Chai-1r, an updated version of Chai-1 with full support for restraints. Prompting Chai-1r with prior structural knowledge can boost accuracy by >2x. https://t.co/aCYtHl207n
1
53
240
Working with this team has been one of the great privileges of my academic career thus far. They’re incredible mentors!
Are you a PhD student interested in working at the forefront of AI and biology? Join me, @KevinKaichuang, @avapamini, @lorin_crawford and Kristen Severson as a research intern in the BioML team this summer! https://t.co/fsba17g5f6 Some previous intern projects in🧵below...
0
1
9
We've released a new version (v0.2.0) of the chai-lab Python package that makes it significantly easier to pass Multiple Sequence Alignments (MSAs) to the Chai-1 model. When provided, MSAs can improve the accuracy of predicted structures. https://t.co/EFz6edaTpf
1
9
31
A good test for any structure prediction model is its predictive accuracy for novel entrants to the PDB. I just ran Chai-1 on structure 9CBK released yesterday on the PDB (and importantly not yet included in the training set for Chai-1). 🖥️🧬 Chai-1 predicted 9CBK's
5
33
223
Thanks to everyone for using our model! Excited to keep making it better and easier for for everyone to use.
We deployed the following improvements to the Chai-1 lab today based on community feedback: - Support for PTMs, by using CCD codes - Email updates for completed jobs - Confidence scores (eg plddt) saved to PDB files - Faster computations
0
0
6
We’re excited to introduce @ChaiDiscovery and release Chai-1, a foundation model for molecular structure prediction that performs at the state-of-the-art across a variety of drug discovery tasks We're releasing inference code, weights & a web interface: https://t.co/QmpbVO9Fhd
70
412
2K
A CLIP-style contrastive model between protein sequences and text descriptions of their function. @Kevin_E_Wu @james_y_zou
4
15
130
Available today: the final journal version of FoldingDiffusion in @NatureComms Led by former intern @Kevin_E_Wu with @alexijielu @vdbergrianne @SarahAlamdari @james_y_zou and @avapamini
We present Folding Diffusion: a diffusion model for protein structure inspired by physical protein folding. Let by @Kevin_E_Wu during his internship, with @alexijielu @vdbergrianne @james_y_zou @avapamini Code: https://t.co/ldKQGnsrMQ Preprint: https://t.co/fGdHqLTi5U
7
59
315
Today, we announce a breakthrough in AI drug design: we are the first to design AND validate new therapeutic antibodies with zero-shot #generativeAI. De novo antibodies are here! Read our preprint: https://t.co/3eI1N8FGfr
#JPM23
2
47
194
Combining DDPMs with Transformers has been done in other areas too!! Really cool work by @Kevin_E_Wu @KevinKaichuang @avapamini et al. combines the two to generate plausible protein structures. It diffuses over dihedral & bond angles Paper:
arxiv.org
The ability to computationally generate novel yet physically foldable protein structures could lead to new biological discoveries and new treatments targeting yet incurable diseases. Despite...
Transformers are all the rage today. But neither DALLE nor Stable Diffusion uses Transformer for image generation. Instead, they rely on a 7-year-old, Jurassic-era neural architecture. Why? 🤷🏾♀️⁉️ It’s finally time for Transformer and Diffusion to join forces! Quick 🧵👇:
2
9
45
Join us online at #MLCB2022 on Monday 11/21! https://t.co/UW0g8oBoV1 My students @KyleWSwanson will present #AI of immune escape and @Kevin_E_Wu will show #diffusion to generate proteins. Collaborator @pkgyawali will present more stable feature interpretations. See you there!
0
3
5
Among white students admitted to Harvard, 54% are athletes+legacy+dean's list+faculty/staff children (column 2). Just 10% is regular admission Big athletic school, Harvard...
91
514
4K