iiinova — where your digital self keeps evolving

Kieran Didi @kdidi.bsky.social · 18/09/2026

We are continuing to push Proteina-Complexa and similar models and collaborating with the research community to show the impact of these models on how science can be done, stay tuned and reach out if you have interesting problems you want to tackle! 5/5

010

Kieran Didi @kdidi.bsky.social · 18/09/2026

It has also been exciting to see Protein-Complexa used in emerging agentic biology workflows—from work at Anthropic and Muni Bio to Levin Harness and the BioNeMo Agent Toolkit (and many others!) SOTA open models can become infrastructure for accelerating science! 4/n

130

Kieran Didi @kdidi.bsky.social · 18/09/2026

With colleagues at Cambridge, we tested our carbohydrate binders in cell-agglutination assays, closer to their intended use case. Several binders showed specificity for A antigen, B antigen, or both - all of them useful behavior for the application. 3/n

110

Kieran Didi @kdidi.bsky.social · 18/09/2026

Sampling temperature gives us a useful dial: low temperature produces stable codesigns, higher temperature yields native-like structures. With collaborators in Munich/Leipzig, we pushed this to an 800-aa codesign—and obtained a crystal structure that closely matches the design! 2/n

130

Kieran Didi @kdidi.bsky.social · 18/09/2026

Our wet-lab validation campaign for Proteina-Complexa is now on bioRxiv! It includes some new exciting experimental results, from large protein structures (fully codesigned!) to functional carbohydrate binders. biorxiv.org/content/10.6... Thread with some of the additions 🧵1/n

biorxiv.org

Latent generative search unlocks de novo design of untapped biomolecular interactions at scale

De novo protein design has advanced rapidly, yet designing binders to polar, solvent-exposed epitopes and small, flexible ligands remains challenging. Such hydrated surfaces and flexible molecules, in...

1145

Kieran Didi @kdidi.bsky.social · 01/08/2026

Fast MSAs are coming to nucleic acids! Riboseek has been super helpful for some of my own work, very excited to see what people do with this cool new tool! Amazing collaboration as always with @milot.bsky.social @martinsteinegger.bsky.social and co, led by @sukhwanpark.bsky.social!

072

Reposted by Kieran Didi

Scott Soderling @scottsoderling.bsky.social · 18/03/2026

My lab was lucky to be able to help test Proteina-Complexa binders experimentally. We were blown away by the results. Congrats to our colleagues at NVIDIA!

061

Reposted by Kieran Didi

bioRxiv Bioinfo @biorxiv-bioinfo.bsky.social · 16/03/2026

Efficient protein structure prediction fromcompact computers to datacenters withOpenFold-TRT www.biorxiv.org/content/10.64898/20…

0127

Kieran Didi @kdidi.bsky.social · 17/03/2026

Also checkout the great post from @karstenkreis.bsky.social about our work! bsky.app/profile/kars...

000

Kieran Didi @kdidi.bsky.social · 17/03/2026

That's it from me! Happy to hear what people think once they try it out (code weights etc all available under permissive license!) 19/19

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

The whole Proteina effort started more than 2 years ago when Arash Vahdat believed in me and my two partners in crime Tomas Geffner and @karstenkreis.bsky.social to start this crazy protein thing at NVIDIA, unreal to see how far we have come since then as a field. 18/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Case study 5/5: first fully de novo Carbohydrate binders We target blood-group B antigen and got a a 21% hit rate. Massive shoutout to my long-term enzyme collaborator @mattpenner.bsky.social at Cambridge for powering this through, excited about what codesign will unlock for enzymes! 17/n

110

Kieran Didi @kdidi.bsky.social · 17/03/2026

Case study 4/5: Nipah virus In the Adaptyv binder competition setting, Proteina-Complexa produced a nanomolar binder (56 nM) to NiV-G, targeting a recessed receptor-binding site. We also show how to go beyond partial diffusion and perform joint sequence-structure redesign 16/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Case study 3/5: Targeting Kinases (PAK1 + CK1δ) with peptide binders Across mini-protein and short-peptide regimes (<31 aa peptides and 49–74 aa mini-binders), we observed strong hit rates on difficult target sites. 15/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Case study 2/5: ActRIIA, a target for GLP-1 associated muscle wasting We designed de novo binders that block myostatin signaling in cells. Tightest binder reached KD = 36 nM, with functional downstream inhibition. 14/n

110

Kieran Didi @kdidi.bsky.social · 17/03/2026

Case study 1/5: PDGFR We achieved a 63.5% hit rate, with top binders reaching double-digit picomolar affinity (best reported at 93.6 pM). Very cool collab with @novonordisk.bsky.social 13/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

As part of this campaign, we also ran a the biggest head-to-head wetlab comparison of design methods ever. Proteina-Complexa outperformed baselines in this setting. My personal highlight: we are the only method where codesign shines, outperforming MPNN for the first time! 12/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Massive campaign: ~1M designed candidates screened across 127 diverse/challenging targets with all-to-all binding readouts. We solve more than 2/3 of the targets given a very limited compute budget and often quite challenging hotspots and crops 11/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

That’s the method side (come to ICLR in Rio to hear more about at our Oral!). But in silico only goes so far, so now the wet-lab side! This was a major collaboration with @manifoldbio.bsky.social , @vivabiotech, @novonordisk.bsky.social , @CambridgeUniversity, @DukeUniversity, @lmu.de . 10/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Quantitatively, these inference-time scaling strategies outperform prior hallucination-based methods under normalized compute budgets as well as other generative models, setting a new SOTA in in-silico binder design. 9/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

One of my highlights: because search is reward-guided, we can optimize for biophysical objectives during generation - including interface hydrogen-bond terms that promote denser interaction networks. With MLFF improvements being rapid, this becomes more and more powerful! 8/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Data was also key: with @martinsteinegger.bsky.social and @sooyoung-cha.bsky.social lab we introduce Teddymer, a large synthetic binder-target pretraining resource built from domain-domain interactions derived from AFDB monomers (details + links on the project page). 7/n

120

Kieran Didi @kdidi.bsky.social · 17/03/2026

In practice, this means scaling inference-time compute using strategies like beam search, MCTS, and Feynman–Kac steering to improve candidate quality and physical plausibility. Force field metrics, interaction energies, folding scores, you choose you reward! 6/n

120

Kieran Didi @kdidi.bsky.social · 17/03/2026

We then combine the strengths of generative models with optimisation methods like hallucination. We call the inference recipe latent generative search: combine a learned generative prior with search at test time to steer toward better binders. 5/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Core design choice: - joint sequence-structure generation in a partially latent atomistic framework (building on La-Proteina). - No discrete sequence tokenization loop. - No mandatory post hoc inverse-folding redesign step, the first codesign model that outperforms MPNN! 4/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Proteina-Complexa is a generative binder design framework that supports diverse targets: single-chain proteins, multi-chain complexes, and small-molecule binding contexts. 3/n

100

Kieran Didi @kdidi.bsky.social · 17/03/2026

Two papers, one story: 1) ICLR 2026 Oral: method + core modeling advances 2) Experimental validation: large-scale wet-lab evidence across many targets and campaigns Let’s start with the method paper. 2/n

101

Kieran Didi @kdidi.bsky.social · 17/03/2026

📢 We’re launching Proteina-Complexa — and after the Jensen keynote mention, we definitely had to post this thread now ;) Atomistic binder design with generative pretraining + test-time compute, plus large-scale wet-lab validation. Project page: research.nvidia.com/labs/genair/... 🧵 1/n

13716

Kieran Didi @kdidi.bsky.social · 12/02/2026

Too many REPA / RAE / representation alignment papers lately? I was lost too, so I wrote a blog post that organizes the space into phases and zooms in on what actually matters for general/molecular ML. Curious what folks think - link below! 🔗 Blog: kdidi.netlify.app/blog/ml/2025...

kdidi.netlify.app

The unification of representation learning and generative modelling

A deep dive into the convergence of discriminative and generative AI, covering 4 phases of evolution from REPA to RAE and beyond.

0112

Reposted by Kieran Didi

Martin Steinegger 🇺🇦 @martinsteinegger.bsky.social · 21/09/2025

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

nature.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...

417463

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

010

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

110

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

100

Kieran Didi @kdidi.bsky.social · 15/08/2025

`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

121

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

111

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

100

Kieran Didi @kdidi.bsky.social · 15/08/2025

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

1215

Reposted by Kieran Didi

Chaitanya K. Joshi @chaitjo.bsky.social · 15/08/2025

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...

05418

Kieran Didi @kdidi.bsky.social · 19/07/2025

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!

research.nvidia.com

La-Proteina: Atomistic Protein Generation via Partially Latent Flow Matching

0249

Kieran Didi @kdidi.bsky.social · 23/04/2025

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

0122

Kieran Didi @kdidi.bsky.social · 04/03/2025

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!

051

Reposted by Kieran Didi

Anshul Kundaje @anshulkundaje.bsky.social · 20/02/2025

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

0122

Kieran Didi @kdidi.bsky.social · 20/02/2025

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

020

Reposted by Kieran Didi

Kresten Lindorff-Larsen @lindorfflarsen.bsky.social · 15/01/2025

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."

0196

Reposted by Kieran Didi

Arne Schneuing @rne.bsky.social · 15/01/2025

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

16425

Kieran Didi @kdidi.bsky.social · 23/12/2024

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

020