iiinova — where your digital self keeps evolving

Reposted by Diego del Alamo

Rasmus Jensen @rasmusjensen.bsky.social · 25/09/2026

🎉 Out today in Cell: the paper I've been looking forward to sharing for a long time 🎉 We used cryo-ET to find a molecular machine nobody knew existed on the surface of a minimal bacterium, and worked out what it does. 🧵(1/7) #TeamTomo #cryoET

1023696 236

Diego del Alamo @delalamo.xyz · 21/09/2026

The preprint for Structure-based Antibody Renumbering is now on biorxiv. The method, which I developed while I was between jobs last year, gives ANARCI-like annotations to antibodies and related systems using only the 3d coordinates of main chain atoms. www.biorxiv.org/content/10.6...

biorxiv.org

Structure-based Antibody Renumbering

Antibody numbering schemes like IMGT and Chothia assign each residue in the variable domain a consistent index based on substructural position. These annotations standardize sequences with different l...

032 3

Diego del Alamo @delalamo.xyz · 09/09/2026

Can you go into specifics?

210 2

Diego del Alamo @delalamo.xyz · 09/09/2026

Folks who’ve been in academia 5+ years: have you tried pointing an LLM at your (presumably huge) folder of random unpublished results, crystal/EM structures, benchmarks, etc, and asked it to craft a story from those scattered data? If so, what did it come up with?

230 3

Diego del Alamo @delalamo.xyz · 04/09/2026

Finally submitted the technical report describing this method for internal review, with any luck it will appear on biorxiv sometime in the next couple weeks. If you work with antibody structure bioinformatics, give it a spin and let me know how it does!

github.com

GitHub - delalamo/SAbR: Structure-based antibody renumbering

Structure-based antibody renumbering. Contribute to delalamo/SAbR development by creating an account on GitHub.

180 8

Diego del Alamo @delalamo.xyz · 09/06/2026

It’s representations of just the CDRs. The FWR is omitted from mean pooling. I’m basically just running a different version of the “junkyard” experiment from the paper www.nature.com/articles/s41... . They should be completely different

110 1

Diego del Alamo @delalamo.xyz · 09/06/2026

The training details for VHHBERT are light, but seems like there is no clustering at all. Anyway, the finding suggests that these models all represent CDRH3 residues as a linear combination of residue type features and position features, which is incredibly underpowered

010 1

Diego del Alamo @delalamo.xyz · 09/06/2026

Despite the huge amount of training data, ESM-C is still unable to distinguish mean-pooled CDRH3 representations of real antibodies from those of antibodies w/ scrambled CDRs. Only VHHBERT seems to do this, and only for natural sequences

270 7

Diego del Alamo @delalamo.xyz · 17/05/2026

This should replace CAPTCHAs on biorxiv

0141 14

Reposted by Diego del Alamo

Pavel @spavel.bsky.social · 15/05/2026

But they were all of them deceived, for another leche was made...

432116397 2.1K

Diego del Alamo @delalamo.xyz · 16/05/2026

A thousand percent, and it makes the short email retention policies at most companies extremely annoying

100 1

Diego del Alamo @delalamo.xyz · 04/05/2026

LinkedIn as an employer looking for candidates is like ipTM: great for filtering candidates discovered elsewhere, but absolute dogshit for discovery. Quite a few people find ways to fluff up mostly-empty CVs and post a firehose of AI-generated posts. Anyway, role still open

182 8

Reposted by Diego del Alamo

Theo Sanderson @theo.io · 23/04/2026

Indeed.. here is "generate the first page of a great Nature paper that will get lots of citations. It must be very novel."

33716 37

Diego del Alamo @delalamo.xyz · 23/04/2026

Takeda Pharma is hiring scientists at both the straight-out-of-PhD and senior levels in the AIML group, particularly those with experience training, modifying, and applying foundation models. Please reach out if youre interested and want to make the jump to industry, link below👇

170 7

Diego del Alamo @delalamo.xyz · 29/03/2026

Thanks! I had it filed under a different paper by Johnson et al 2024 that had nothing to do with protein fitness

010 1

Diego del Alamo @delalamo.xyz · 29/03/2026

Does anyone know what paper this figure is from? The citation I have for it turned out to be wrong

120 2

Diego del Alamo @delalamo.xyz · 26/03/2026

PDB 7E5Y is a funny case of two copies of the same protein in the same asymmetric unit, one of which has a register shift error (chain C, green) while the other does not (chain H, orange). Here's conserved J-gene Trp 118 (IMGT; 108 in PDB). Trastuzumab shown in teal for reference

020 2

Diego del Alamo @delalamo.xyz · 25/03/2026

Bits to binders* not bits in bio

010 1

Diego del Alamo @delalamo.xyz · 25/03/2026

The proteina-complexa paper is the 2nd this month suggesting that high K+E content, a staple of ProteinMPNN designs, is predictive of poor expression (the bits in bio benchmark paper showed a few weeks ago). In their case, they measure poor sequence recovery in phage display

2234 23

Diego del Alamo @delalamo.xyz · 18/03/2026

This is new? Instead of relying on intermediate estimates of the denoised state, Proteina-Complexa just does the full denoising, calculates whatever it needs to, and sends the info back to continue guiding the unfinished diffusion roll-out

130 3

Diego del Alamo @delalamo.xyz · 18/03/2026

Sometimes you hit a home run, and sometimes you get to first base

010 1

Diego del Alamo @delalamo.xyz · 17/03/2026

What? In like two days? How on earth?

131 3

Diego del Alamo @delalamo.xyz · 15/03/2026

So I can't say I've ever seen residual cross-attention before (where the final representations attend to earlier representations of the input data); is there any literature on when and where to use this?

070 7

Reposted by Diego del Alamo

jproney @jproney.bsky.social · 13/03/2026

Diffusion models actually learn a series of time-indexed energy landscapes, which are corrupted with different amounts of noise. The ranking ability of ProteinEBM peaked slightly above t=0. Inspired by this finding we trained an "expert" model only on low time levels, which we call ProteinEBM-x.

151 5

Diego del Alamo @delalamo.xyz · 13/03/2026

You can triple your hit rate without changing your NN by simply using a different search algorithm during sampling

020 2

Diego del Alamo @delalamo.xyz · 13/03/2026

Thanks for tagging! Very interesting read…

010 1

Reposted by Diego del Alamo

Kaitlin @justktln2.bsky.social · 11/03/2026

> paper claims astonishing progress on protein folding problem > ask if it’s interesting proteins or villin headpiece denatured with urea > they say results are robust to protein sequence properties > open the pdf > villin headpiece in urea

181 8

Reposted by Diego del Alamo

Joe Greener @jgreener64.bsky.social · 15/10/2025

I wrote a blog post about the future of structural bioinformatics. Where to go after AlphaFold? How do we avoid the field becoming a load of half-baked LLMs? Let me know what you think. jgreener64.github.io/posts/struct...

jgreener64.github.io

Where next for structural bioinformatics?

45316 53

Diego del Alamo @delalamo.xyz · 10/03/2026

One of the most interesting parts of this workflow is the "sunk cost fallacy" estimator that predicts how promising a particular mutational line of inquiry is, and whether it is worth abandoning in favor of others

020 2

Diego del Alamo @delalamo.xyz · 08/03/2026

Most benchmarks for drug discovery AI don't effectively evaluate generative models; instead, due to the data's incompleteness, they rely on surrogate functions, like fwd folding, property prediction, or ranking previously characterized designs. No idea what the solution is here

030 3

Reposted by Diego del Alamo

Kresten Lindorff-Larsen @lindorfflarsen.bsky.social · 08/03/2026

New paper from former PhD student @tkschulze.bsky.social on supervised learning of protein variant effects across large-scale mutagenesis datasets MAVE/DMS experiments provide large amounts of data for benchmarking variant effect predictors, but may be difficult to use in supervised learning. 1/5

1229 22

Reposted by Diego del Alamo

Ethan @ethanrosenthal.com · 05/03/2026

you and me both claude

0252 25

Diego del Alamo @delalamo.xyz · 05/03/2026

I regret to inform you that if you are job hunting on LI and this is your profile pic then you’re ngmi

030 3

Diego del Alamo @delalamo.xyz · 05/03/2026

Somewhere a student is wondering why their stats lecture on Gaussian Processes has so many pictures of the Strait of Hormuz

010 1

Diego del Alamo @delalamo.xyz · 04/03/2026

It’s tough out there! I just feel bad because I’m not the hiring manager, just RTing openings to get visibility

110 1

Diego del Alamo @delalamo.xyz · 03/03/2026

RIP my LinkedIn after mentioning a job opening in my group ☠️

2100 10

Diego del Alamo @delalamo.xyz · 03/03/2026

The two co-first authors of this research paper, Yang & Yang, has decided to sort their names alphabetically

041 4

Diego del Alamo @delalamo.xyz · 28/02/2026

Protenix trained an identical model with way more training data (2025 cutoff instead of 2021), demonstrating that antibody-antigen modeling, but not protein-ligand modeling, is currently data-limited (DQ SR % means % DockQ≥0.23) with this architecture

3100 10

Diego del Alamo @delalamo.xyz · 27/02/2026

Extremely important point from the authors! The entire pipeline *is* the method. Not the NN per se! My bad for misjudging

010 1

Diego del Alamo @delalamo.xyz · 27/02/2026

Ah! I understand now. Let me add a note to what I wrote on the other site

120 2

Diego del Alamo @delalamo.xyz · 27/02/2026

Thanks, I appreciate your feedback, and I see how this decision makes some sense. But don't you think it is misleading to say that it is a problem of the method, rather than how it is applied? They might improve w/ different masking (e.g., using pseudoperplexity instead of marginal likelihoods)

110 1

Diego del Alamo @delalamo.xyz · 26/02/2026

Scribbled on at least one whiteboard in every office with "do not erase" written below

010 1

Diego del Alamo @delalamo.xyz · 26/02/2026

First fig below is the relevant code for zero-shot fitness prediction w/ ESM in the ProteinGym repo for example. It would explain why the results in the preprint, second fig, have many masked sequence-only models near the bottom of the rankings (3/3)

000 0

Diego del Alamo @delalamo.xyz · 26/02/2026

As far as I can tell they are using precomputed ProteinGym predictions. For masked LMs, ProteinGym uses a marginal perplexity which adds together the individual logits for all mutations against a WT or partially masked background, which definitionally cant predict epistasis (2/3)

100 1

Diego del Alamo @delalamo.xyz · 26/02/2026

I have a concern with this paper and I want someone who knows more than me to confirm if it is founded or not. The title makes a pretty specific claim about epistasis predictions, but the method does not seem sound for masked LMs (1/3)

310 3

Diego del Alamo @delalamo.xyz · 26/02/2026

Hot take: this "chain X[auth Y]" notation on FASTA files pulled from the PDB is needlessly confusing, adds nothing, and needed to be changed yesterday

270 7