Lightnews — Scholar-powered news

Reposted by Kieran Didi

Nature Methods @natmethods.nature.com · 20d

GPU-accelerated MMseqs2 offers tremendous speedup for homology retrieval, protein structure prediction with ColabFold, and protein structure search with Foldseek. @martinsteinegger.bsky.social @milot.bsky.social @machine.learning.bio

www.nature.com/articles/s41...

GPU-accelerated homology search with MMseqs2 - Nature Methods

Graphics processing unit-accelerated MMseqs2 offers tremendous speedups for homology retrieval from metagenomic databases, query-centered multiple sequence alignment generation for structure predictio...

www.nature.com

21 81

Reposted by Kieran Didi

Martin Steinegger 🇺🇦 @martinsteinegger.bsky.social · 17d

MMseqs2-GPU sets new standards in single query search speed, allows near instant search of big databases, scales to multiple GPUs and is fast beyond VRAM. It enables ColabFold MSA generation in seconds and sub-second Foldseek search against AFDB50. 1/n
📄 www.nature.com/articles/s41...
💿 mmseqs.com

GPU-accelerated homology search with MMseqs2 - Nature Methods

Graphics processing unit-accelerated MMseqs2 offers tremendous speedups for homology retrieval from metagenomic databases, query-centered multiple sequence alignment generation for structure predictio...

www.nature.com

4 64 170

Kieran Didi @kdidi.bsky.social · Aug 15

For more details read the thread by the man himself: bsky.app/profile/ncor...

Nate Corley @ncorley.bsky.social · Aug 15

(1/7)
Training biomolecular foundation models shouldn't be so hard. And open-source structure prediction is important. So today we're releasing two software packages: AtomWorks and RosettaFold3 (RF3)

[https://www.biorxiv.org/content/10.1101/2025.08.14.670328v2](www.biorxiv.org/content/10.1...)

Accelerating Biomolecular Modeling with AtomWorks and RF3

Deep learning methods trained on protein structure databases have revolutionized biomolecular structure prediction, but developing and training new models remains a considerable challenge. To facilita...

www.biorxiv.org

1

Kieran Didi @kdidi.bsky.social · Aug 15

An incredible project to witness, led by the most incredible dreamteam @ncorley.bsky.social , @simonmathis.bsky.social and Rohith Krishna with an amazing team inside and outside the Baker lab. Check it out and let us know what you think/contribute to the codebase! 6/6

1 1

Kieran Didi @kdidi.bsky.social · Aug 15

The preprint shows how atomworks leads to better reference conformers (and better predictions!), enables advanced features in RF3 like chirality-aware training or ligand templating and narrows the performance gap to closed-source models. 5/6

1

Kieran Didi @kdidi.bsky.social · Aug 15

`atomworks.ml` on the other hand offers advanced dataset featurization and sampling for deep learning workflows, all operating on the canonical AtomArray object from @biotite_python so that all transforms are traceable and generalizable between models. 4/6

1 1 2

Kieran Didi @kdidi.bsky.social · Aug 15

AtomWorks has two main components: atomworks.io takes a file (cif, sdf, ...) and does parsing, cleaning and more. You can also look at your structures in a notebook or via PyMol thanks to pymol-remote, so you can directly inspect if your code does what you want! 3/6

1 1 1

Kieran Didi @kdidi.bsky.social · Aug 15

In the past, every BioML model had its own data pipeline, creating loads of overhead. With AtomWorks, >80% of code is shared between models like ProteinMPNN, RF3 or design models. 2/6

1

Kieran Didi @kdidi.bsky.social · Aug 15

AtomWorks is out! Building upon @biotite_python, we built a toolkit for all things biomolecules and trained RF3 with it. All open-source, test it via `pip install atomworks`!

AtomWorks: github.com/RosettaCommo...
RF3: github.com/RosettaCommo...
Paper: tinyurl.com/y2w4z65b

1/6

1 6 21

Reposted by Kieran Didi

Chaitanya K. Joshi @chaitjo.bsky.social · Aug 15

RosettaFold 3 is here! 🧬🚀

AtomWorks (the foundational data pipeline powering it) is perhaps the really most exciting part of this release!

Congratulations @simonmathis.bsky.social and team!!! ❤️

bioRxiv preprint: www.biorxiv.org/content/10.1...

19 52

Reposted by Kieran Didi

Nate Corley @ncorley.bsky.social · Aug 15

(1/7)
Training biomolecular foundation models shouldn't be so hard. And open-source structure prediction is important. So today we're releasing two software packages: AtomWorks and RosettaFold3 (RF3)

[https://www.biorxiv.org/content/10.1101/2025.08.14.670328v2](www.biorxiv.org/content/10.1...)

Accelerating Biomolecular Modeling with AtomWorks and RF3

Deep learning methods trained on protein structure databases have revolutionized biomolecular structure prediction, but developing and training new models remains a considerable challenge. To facilita...

www.biorxiv.org

2 28 67

Kieran Didi @kdidi.bsky.social · Jul 19

Very excited about our latest all-atom generative model proteina, check out the project page (research.nvidia.com/labs/genair/...) and stay tuned for the code release soon!

9 23

Kieran Didi @kdidi.bsky.social · Apr 23

Excited to present my first paper officially as a PhD student now as an ICLR Oral this week! Super fun work with the GenAIR team at NVIDIA.

Talk: Fr 10:54 - 11:06 (Oral Session 3B, Garnet 213-215)
Poster: Fr 15:00-17:30 (Hall 3 + Hall 2B #5)

Come by the poster/reach out to chat

2 11

Kieran Didi @kdidi.bsky.social · Mar 4

Such a fun project to work on with a stellar team! Stay tuned for other things to come here, and see you all in Singapore!

Karsten Kreis @karstenkreis.bsky.social · Mar 4

📢📢 "Proteina: Scaling Flow-based Protein Structure Generative Models"

#ICLR2025 (Oral Presentation)

🔥 Project page: research.nvidia.com/labs/genair/...
📜 Paper: arxiv.org/abs/2503.00710
🛠️ Code and weights: github.com/NVIDIA-Digit...

🧵Details in thread...

(1/n)

1 5

Reposted by Kieran Didi

Anshul Kundaje @anshulkundaje.bsky.social · Feb 20

Yet another story of issues with benchmarks and evaluations in ML4bio + a much stronger and fair benchmark #bioMLeval

briantrippe.bsky.social @briantrippe.bsky.social · Feb 19

🔥 Benchmark Alert! MotifBench sets a new standard for evaluating protein design methods in motif scaffolding.
Why does this matter? Reproducibility & fair comparison have been lacking—until now.
Paper: arxiv.org/abs/2502.12479 | Repo: github.com/blt2114/Moti...
A thread ⬇️

2 12

Kieran Didi @kdidi.bsky.social · Feb 20

Have a look at our shiny new benchmark for motif-scaffolding in computational protein design! New (and harder) tasks, including a reproducible evaluation pipeline

briantrippe.bsky.social @briantrippe.bsky.social · Feb 19

🔥 Benchmark Alert! MotifBench sets a new standard for evaluating protein design methods in motif scaffolding.
Why does this matter? Reproducibility & fair comparison have been lacking—until now.
Paper: arxiv.org/abs/2502.12479 | Repo: github.com/blt2114/Moti...
A thread ⬇️

2

Reposted by Kieran Didi

Kresten Lindorff-Larsen @lindorfflarsen.bsky.social · Jan 15

This!

Also well put in this editorial in PLOS Comp Biol:
Putting benchmarks in their rightful place: The heart of computational biology
doi.org/10.1371/jour...

Screenshot from paper that says:
"Developing good and comprehensive benchmarks, in which the performance metrics of each tool reflect its real-world utility, requires a significant effort. For highly competitive and established fields, such as protein structure predictions, community experiments evaluating the methods have been held periodically to provide blinded assessments of prediction performance. These blinded assessments are perhaps the gold standard on how benchmarks should be run. However, in most areas of computational biology, no such regular blinded contests are available. Instead, many tool developers end up generating their own benchmarks, which they publish alongside a newly developed tool to show its improved performance. The downside of this approach is that, if a new approach is developed in parallel to assembly of the benchmark on which it is evaluated, there is a strong selection bias encouraging the authors to report tool development approaches performing well against the benchmark compared to previous tools. This reporting bias makes most benchmarks that accompany newly developed tools questionable. Even if the authors are aware of this problem and take conscious steps to separate benchmark, evaluation method, and method development, subconscious bias may persist and affect the final outcome."

6 19

Reposted by Kieran Didi

Arne Schneuing @rne.bsky.social · Jan 15

Our paper on computational design of chemically induced protein interactions is out in @natureportfolio.bsky.social. Big thanks to all co-authors, especially Anthony Marchand, Stephen Buckley and Bruno Correia!

t.co/vtYlhi8aQm

1 25 65

Kieran Didi @kdidi.bsky.social · Dec 23

So excited for you and what is to come! Onwards and Upwards;)

2

Reposted by Kieran Didi

Michael Bronstein @mmbronstein.bsky.social · Dec 9

After two years, our paper on generative models for structure-based drug design is finally out in @natcomputsci.bsky.social

www.nature.com/articles/s43...

Structure-based drug design with equivariant diffusion models - Nature Computational Science

This work applies diffusion models to conditional molecule generation and shows how they can be used to tackle various structure-based drug design problems

www.nature.com

2 38 170

Reposted by Kieran Didi

Rory Byrne @rory.bio · Dec 3

Thanks Jascha 🫶

We’re working hard to create sustainable funding mechanisms for open source scientific tooling - understanding the challenge landscape is a key first step!

Jascha Achterberg @achterbrain.bsky.social · Dec 3

Rory @rory.bio is working on a large project to accelerate science 🧪 through high-quality & open-source software.

To steer the project it would be amazing to hear from scientists across fields about problems in the scientific process *you* want to see solved! Tell Rory here flywhl-ideas.notion.site

Notion – The all-in-one workspace for your notes, tasks, wikis, and databases.

A new tool that blends your everyday work apps into one. It's the all-in-one workspace for you and your team

flywhl-ideas.notion.site

2 9

Kieran Didi @kdidi.bsky.social · Nov 29

Love PyMOL Remote, one of these tools that does one thing and does it well!

Simon Mathis @simonmathis.bsky.social · Nov 29

For some more guidance on how to use this, Martin Buttenschön wrote a nice blogpost: www.blopig.com/blog/2024/11...

1 7

Reposted by Kieran Didi

Kresten Lindorff-Larsen @lindorfflarsen.bsky.social · Nov 27

This should go chiral

estella newcombe @estellaan.bsky.social · Nov 27

We did this crazy project where we tried to see if proteins could interact with their mirror image ligand. Seems impossible when proteins need to form 3D structures to interact. But what about if the interaction remains disordered???

www.nature.com/articles/s41...

Stereochemistry in the disorder–order continuum of protein interactions - Nature

Studies on protein–protein interactions using proteins containing d- or l-amino acids show that stereoselectivity of binding varies with the degree of disorder within the complex.

www.nature.com

7 17 150

Kieran Didi @kdidi.bsky.social · Nov 16

MSAs go brrr with MMseqs2-GPU! Super fun project, happy to work with and learn from a stellar team of engineers and scientists. Try it out and stay tuned!

📄 Preprint: www.biorxiv.org/content/10.1...
💾 Code: mmseqs.com
🗞️ Blog: developer.nvidia.com/blog/boost-a...

7 29