Ramith Hettiarachchi

PhD Student @CMUPittCompBio / @SCSatCMU

Interested in ML for science while striving for trustworthiness

[Ramith Hettiarachchi @ramith__]

We’ve been making phylogenetic trees differentiable :)

Check out our work at #ICML2023 workshops - Sampling and Optimization in Discrete Space (SODS) ፨ and Differentiable Almost Everything (DiffAE) 〆

Looking forward to discuss and learn more! 🙌
❤️ work with Avi & @sokrypton

[Ramith Hettiarachchi @ramith__]

@ravinhwg decent so far.. from the UX side could improve (long running tasks, do not share much info with the user until its done/unless asked)

[Ramith Hettiarachchi @ramith__]

gemini carbon is decent.. doesn't freak out when it hears `Atomic Motif Enzyme benchmark` 👌

[Ramith Hettiarachchi @ramith__]

@lpachter @3blue1brown @NarekMaloyan oh just saw the export options.. really cool!

[Ramith Hettiarachchi @ramith__]

did a group presentation today with @3blue1brown's manim style transitions. Sharing the template for anyone to build upon

Brought to life with @NarekMaloyan's manim-web + reveal.js + claude + codex

view template: https://t.co/LOSKmrtlXa
view repo: https://t.co/gTb9xxYp7F

[Ramith Hettiarachchi @ramith__]

@lpachter @3blue1brown @NarekMaloyan wow! just installed, looks pretty cool!

I wonder if manim-web could be another backend for the tool...

The friction I had with manim-slides was, when exporting, things get saved as a video. But I assume manimble can present through python natively so can iterate faster?

[Lior Pachter @lpachter]

@ramith__ @3blue1brown @NarekMaloyan You might find Manimble useful (for editing manim slides). https://t.co/IPOZvYsA1Z

[Ramith Hettiarachchi @ramith__]

@MuzafferKal_ 😆 is this 5.5 or 5?

[Sergey Ovchinnikov @sokrypton]

https://t.co/7OlBz2wKP1 updated to include:
- bioemu2 @FrankNoeBerlin
- mol1 @MolenceBio
- fastplm (confidence modules for esmfold fast models) @Logan_Hallee
- af3x crosslinks @jankosinski
- faster diffusion (exponential update and geometric noise schedule) @vinay_matt

[Ramith Hettiarachchi @ramith__]

❤️

[Sashenka Gamage @sash__2001]

really liked ramith aiya’s idea to do this, so wanted to try something fancy as well.

i was reading about world models and decided to make a slide deck for @NVIDIARobotics DreamZero — paper “World Action Model are Zero-shot Policies”

https://t.co/TcyFYYsHlI

[Mark Dredze @mdredze]

This. 100% this.

When I use AI to make slides, my first step is to figure out what I want to say, how I want to say it, and what visuals to use to convey that message.

Only then do I use AI to assemble the slides. I then edit heavily.

[Garyk Brixi @garykbrixi]

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 🧵

[Yehlin Cho @ChoYehlin]

🚀 Excited to share BindClaw! 🦀

Starting from a prompt or research paper, BindClaw connects biological mechanisms to protein design: identify targets and binding sites → build mechanism-specific computational assays → generate and assess binders → refine designs using the results.

We use agents to evaluate whether designs fit the intended biological objective, including competition with a native partner and preference for a particular molecular state. 🧬

In a human-guided calmodulin campaign, 15 of 18 designs showed binding to calcium-loaded CaM, including subnanomolar binders supported by both BLI and SPR.

This work is an incredible team effort together with @DidiKieran , @saint__ai, @ Kyle Gion and @sacdallago and many others!

📄paper/project: https://t.co/MkjU6Swke7
💻code: will be released in a few weeks!

[Ramith Hettiarachchi @ramith__]

“Never let your agents generate complexity faster than your own mental is expanding” — Argon

[Chirath Hettiarachchi @chirathyh]

I’d love your feedback on CAPSML: explore virtual Type 1 Diabetes scenarios and compare glucose-control strategies without writing code.

Researchers, educators and anyone curious: give it a try.

For research and education only.
https://t.co/6ZmheHRpIJ

[Ramith Hettiarachchi @ramith__]

if you are doing an A/B test to measure the data mix’s effect on training, should you reset the optimizer states?

[Chirath Hettiarachchi @chirathyh]

investing time to learn from a cool paper 🔥, @labmlai style by @vpj and team https://t.co/1ibt1D2Cv2.

worth integrating into my dev workflow — using AI to implement/move faster while still investing time to genuinely understand the underlying ideas.

[Ramith Hettiarachchi @ramith__]

@NarekMaloyan Thanks!

[Ramith Hettiarachchi @ramith__]

Is there a optimized way to run AF3 on tpus? (Comparable to gpu speeds)

[bioZone @thebiozone]

Colchicine is a toxic alkaloid found in some lily species, however this deadly poison can also be used in molecular biology labs or as a medicine for gout

watch on YouTube to support us 👇
https://t.co/Wopsu75yVv