James Lingford

Journals that published Richard Lynn's racist 'research' articles should retract them Richard Lynn's work has been repeatedly condemned for using flawed methodology and deceptively collated data to support racism. It's past time to retract the studies.

The Elsevier investigation mentioned was initiated after a STAT news piece on the problem, co-authored with Cathryn, @dsamorod.bsky.social, @stairwaytokevin.bsky.social, @jedidiahcarlson.com, @jameslingford.bsky.social and @jowiph.bsky.social

www.statnews.com/2024/06/20/r...

Super cool talk from @kmichie.bsky.social. Lots of examples of how AlphaFold3 does weird stuff

Excited to release BoltzGen which brings SOTA folding performance to binder design! The best part of this project is collaborating with a broad network of leading wetlabs that test BoltzGen at an unprecedented scale, showing success on many novel targets and pushing the model to its limits!

As much as I like making plots with matplotlib and seaborn, the way repositioning legends/colourbars works is super unintuitive and a massive timesink.

Viro3D: a comprehensive database of virus protein structure predictions | Molecular Systems Biology imageimageViro3D provides proteome-level, high confidence AI-protein structure predictions for >4,400 viruses, allowing mapping of form and function across the human and animal virosphere. Viro3D i...

Viro3D paper is out! We predicted 85,000 protein structures from human & animal viruses. 1/5 🧵

📑 Paper doi.org/10.1038/s443...
🔭 Explore virosphere viro3d.cvr.gla.ac.uk

RFdiffusion3 is here: www.biorxiv.org/content/10.1...

Switched from using pandas to @pola.rs and it has been a total life saver. Handling dataframes is way more intuitive with the method chaining syntax. Prettier terminal printing, better error messages, lightning fast, fewer headaches.

Recent advances in the inference of deep viral
evolutionary history journals.asm.org/doi/10.1128/...

A review on structural phylogenetics utility and limitations

Protein Structural Phylogenetics Abstract. Protein structural phylogenetics is an interdisciplinary branch of molecular evolution that (i) uses 3D structural data to trace evolutionary his

Fantastic review on methodological developments in protein structural phylogenetics
academic.oup.com/gbe/article/...

Large protein databases reveal structural complementarity and functional locality - Nature Communications Researchers mapped the protein structure landscape, revealing structural complementarity across databases and functional clustering in specific regions. Their web tool helps explore this space, unlock...

Large protein databases reveal structural complementarity and functional locality
www.nature.com/articles/s41...

RFdiffusion2 is now live!
github.com/RosettaCommo...

You can now design proteins, and in particular enzymes from just partially defined amino acid side chains, and without defining their sequence position or order!

GitHub - lehner-lab/combinatorialcores: Source code for analyses and figure reproduction in "Genetics, energetics, and allostery in proteins with randomized cores and surfaces", Escobedo et. al Scienc... Source code for analyses and figure reproduction in "Genetics, energetics, and allostery in proteins with randomized cores and surfaces", Escobedo et. al Science 2025 - lehner-lab/combina...

This is a remarkable paper on protein evolution from the Lehner lab. What makes it especially cool is that all the code to reproduce all the analysis/plots is included in Jupyter notebooks up on Github
www.science.org/doi/10.1126/...
github.com/lehner-lab/c...

• a big concern I have with these LLMs is how they might end up changing our attitudes around productivity. The promise is instant results & faster publications. But the cost will end up being garbage results that are only caught when it's too late

Just to piggyback on this critique of LLMs for coding with some of my own critiques:

• it's a bad idea to install a tool on your machine that has vast read+write permissions
• code that works but is really outputting false info is a real danger and hard to catch. LLMs make that danger worse

Does sequence clustering confound AlphaFold2? Predicting multiple conformational states of proteins represents a significant open challenge in structural biology. Increasingly many methods have be…

Does sequence clustering confound AlphaFold2?
www.sciencedirect.com/science/arti... From @hkws.bsky.social

MMseqs2 v18 is out
- SIMD FW/BW alignment (preprint soon!)
- Sub. Mat. λ calculator by Eric Dawson
- Faster ARM SW by Alexander Nesterovskiy
- MSA-Pairformer’s proximity-based pairing for multimer prediction (www.biorxiv.org/content/10.1...; avail. in ColabFold API)
💾 github.com/soedinglab/M... & 🐍

GitHub - chevrettelab/gator-gc Contribute to chevrettelab/gator-gc development by creating an account on GitHub.

Pretty amazing looking tool for analysing genetic neighbourhoods:
github.com/chevrettelab...

💥 Excited to introduce Bacformer 🦠 - the first foundation model for bacterial genomics. Bacformer represents genomes as sequences of ordered proteins, learning the “grammar” of how genes are arranged, interact and evolve.

Preprint 📝: biorxiv.org/content/10.1...

🧵 1/n

Excited to share our work on WitChi! 🛠️🖥️
We tested it on the GTDB r220 archaeal supermatrix (5,869 taxa & 10,101 cols) removing 55% of sites in <2h.

The phylogeny showed several interesting groupings with overall improved branch support:
#phylogenetics #ArchaeaSky #MSA #opensource #MEvoSky #MicroSky

GitHub - zeqianli/tgv: Explore 5,000+ genomes in the terminal. Light, blazing fast 🚀, vim-motion. Explore 5,000+ genomes in the terminal. Light, blazing fast 🚀, vim-motion. - zeqianli/tgv

New tgv release: local cache!

tgv download hg38

Download UCSC reference genomes to a local sqlite db for much faster browsing. Awesome Rust tools (twobit, bigtools) made this simple.

github.com/zeqianli/tgv

Logo for the Sandpiper website

Out in @natbiotech.nature.com: Metagenome taxonomy profilers usually ignore unknown species. SingleM is an accurate profiler which doesn't, even detecting phyla with no MAGs. Profiles of 700,000 metagenomes at sandpiper.qut.edu.au. A 🧵

Got this setup working where I can now run .ipynb notebooks right from inside the terminal with a combination of neovim, quarto, kitty, and this neovim plugin called molten: github.com/benlubas/mol...
Never have to abandon my precious vim setup again

Inhibiting heme piracy by pathogenic Escherichia coli using de novo-designed proteins - Nature Communications Many pathogens encode transporters that extract heme directly from host proteins. In this study, the authors demonstrate the utility of de novo-designed proteins in understanding the mechanism behind ...

Excited to share our latest work using AI-designed proteins to block heme-piracy by E. coli. Published in @natcomms.nature.com. A team effort between my lab and the ‪‪@knottrna.bsky.social‬ ‬lab, with experimental work led by the talented @danielrfox.bsky.social
www.nature.com/articles/s41...

Been eagerly awaiting this one. Amazing work

We have written up a tutorial on how to run BindCraft, how to prepare your input PDB, how to select hotspots, and various other tips and tricks to get the most out of binder design!

github.com/martinpacesa...

Some matplotlib work in progress

1/27 We have a new paper out! Turns out that snowflake yeast have been hiding a secret from us - they've evolved a (very!) crude circulatory system. Not with blood vessels or a heart, but through spontaneous fluid flows powered by their metabolism. 🧪🔬

www.science.org/doi/full/10....

Closer... I think at this point the solution lies in manually making a list of the hex codes, but that's for another day.

Revisiting this topic now that I've forced myself to use PyMOL. Using this script to install the viridis family of colour palettes: github.com/smsaladi/pym... and running '"spectrum count, palette=magma, MODEL_NAME". The palette is not there still. They must modify the magma palette somehow

How do we know for sure if we have the best AF prediction? We still need prior knowledge/expectations of what we're trying to predict with AF. And in the absence of that, I guess we can't really know unless we try every possible combination in the search space (which is not feasible).