Paul T Kim
@paultkim_ipd
PhD Student, @UWproteindesign - excited to work on machine learning methods for rational protein design. Also core @DNADeviants.
With David and the Baker Lab in the spotlight today, I wanted to share some insights into the @UWproteindesign and how it operates, a glimpse behind the curtain. I had planned to write this post-graduation, but now seems as good a time as any. (Got twitter blue free trial so this…
BREAKING NEWS The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”
The true value of jupyter notebook for ML/data science is that it is a perfect debugging tool. Your project code should be structured in a way that makes it easy to copy paste/import large sections into a jupyter notebook for when you have to debug.
I’m excited to share our significantly-updated preprint on de novo antibody design, where we now demonstrate the structurally accurate design of scFvs (in addition to VHHs) with RFdiffusion! biorxiv.org/content/10.110…
FusOn-pLM is out in @NatureComms!!! 🎉 An INCREDIBLE effort (and a lot of sleepless nights! 🛏️) from my AMAZING 2nd year PhD student @SophieVincoff and her team to train a new pLM that captures properties of some of the MOST disordered and pathogenic proteins known: fusion…
Evolution of the setup - personal laptop running separately w/ linked keyboard and mouse, ipad sidecar for terminal/calendar/todolist, poster board to not get distracted by people walking down the hall, logitech ergonomic keyboard and vertical mouse (saved my wrists)

Basically the customer support experience with any Google produce. Idk what they are doing over there but it seems like anything that cannot be fully automated goes into a doom spiral of low level customer support reps saying they will try to escalate and then nothing happening.
I'm not able to get in touch with anyone except the automated chat, which just sends me a template-like email from a different name every 7 days (Venkatesh, Sai, Praveen, Abdullah). I'm also not eligible to upgrade to a higher support tier.
Great post, cryo is so cool.
A primer on ML in cryo-electron microscopy (cryo-EM) owlposting.com/p/a-primer-on-… confused about cryo-EM??? i explain why people do it, how it works and some ML problems in the area via explanations of 3 @ZhongingAlong papers 7.9k words, 36 minutes reading time. ToC in thread❄️
i rabbit-holed into the Genesis Sim codebase because the website is hypey and unclear; and I didn't want to just blindly retweet. It's basically like Mujoco but with more advanced materials/rendering/solvers, written all in Python thanks to being powered by Taichi, which makes it…
Everything you love about generative models — now powered by real physics! Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics…
this looks unreal
Everything you love about generative models — now powered by real physics! Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics…
Madrid’s metro was 71 miles long in 1995. That would be the world’s 51st longest today, reasonable considering Madrid is the 57th largest city by population. Yet over 12 years, the metro would ~3x in length at costs much lower than was thought possible. Here’s how they did it🧵
Just IPD things ❤️ Another 160 gpus, brings the internal cluster up to 1079 gpus total.

When you call a business and the automated answer goes on about hours, location etc. Very annoying. If I’m calling it’s because I want info that I can’t just look up online.
That deep anxiety when you hit play on the validation run / open wandb and wait for the metrics to load. The universe has decided whether your work was good and now the result is about to be revealed to you.
Join me for the @lab_berger seminar on Wednesday where I'll be talking about #PoET, shrinking protein language models, and protein engineering @openprotein!
This is an ice core from West Antarctica. The black ring on it is a layer of volcanic ash deposited some 21,000 years ago. This and other ice cores can help us to learn about the past, including Phoenician and Roman history🧵
If you are a technical person (engineer, quant, mathematician) interested in learning biology, I encourage you to hit textbooks and memorize the "things", eg. cytokines, surface proteins, cell types, pathways, in core fields like immunology. You will be averse to this because…
1/ Can Large Language Models (LLMs) truly reason? Or are they just sophisticated pattern matchers? In our latest preprint, we explore this key question through a large-scale study of both open-source like Llama, Phi, Gemma, and Mistral and leading closed models, including the…
Now online: We created de novo insulin receptor agonists using computational protein design! @EunheeChoi12 @XiaochenB @PreethamVi @UWproteindesign #BakerLab biorxiv.org/content/10.110…