So, if you want to try it out: Download on @huggingface. Play with the interactive version with @tamarindbio. Study the code on @github. Links to follow.

OpenFold3 is finally out! New fully open-source AF3 reproduction from @open_fold Consortium Try it out on @tamarindbio now! The model is released with training code, similar performance to AlphaFold3 on protein-ligand complexes, best performance ever for RNA structure

BoltzGen: New generalized binder design protocol, with wet lab validation on diverse targets! (From Boltz team+collaborators) On @tamarindbio right now! The authors test only 15 nanobody designs against each of 9 targets. These targets are selected for their high dissimilarity

The Baker Lab designs protein on/off switches 1/🧵 Instead of optimizing for how tightly a given protein binds to a target, Adam and team develop a method to control the duration of binding.

Even better at de novo nanobody design! Germinal from @Stanford and the @arcinstitute finds 4-22% hit rates for AI-designed VHHs against many targets. 4 diverse protein targets (PD-L1, IL-3, IL-20, BHRF1). Screening libraries were 43–101 designs using a split-luciferase triage,

🚨New de novo antibody design protocol! 7/60 hit rate against PD-L1 (IgGM) IgGM is a generative functional antibody design model, supporting de novo design, affinity maturation, inverse folding and more! From 10k candidates over 24 CDR-length grids, 60 were tested; 7 hits

De novo protein binders against undruggable targets? Here's how Institute of Protein Design scientists target Intrinsically Disordered Regions using RFdiffusion 1/🧵 The authors design binders for 43 targets (18 synthetic peptides + 21 diverse native IDRs + a few others,

@tamarindbio now supports deploying internal BioML models as production ready applications to your scientists! Use open-source and internal computational tools, all in one place with a scientist-friendly UI and programmatic API, both at massive scale. We've had our custom tool

(nearly) Everything you need to know for antibody developability & property prediction, a guide At @tamarindbio, we gathered best practices + the most effective (mostly open) in silico tools across all key risks: Aggregation propensity: why it kills programs; sequence +

RFdiffusion2 now available! Atomic-level scaffolding for complex active sites Available on @tamarindbio now. RFdiffusion2 shows state-of-the-art atomic-level scaffolding that generalizes to complex, multi-island active sites, and it reliably yields active enzymes with modest

New Experimentally Validated De Novo Peptide Design Tool, no structure needed! PepMLM is now fully open-source, try it out on @tamarindbio today. 1/🧵

Can't keep up with all the new computational tools? Here are the top ones to know for biology, chemistry and informatics Bioinformatics BLAST MAFFT Foldseek MMseqs Biotite Biopython Biopandas RCSB PDB UniProt InterPro 𝘊𝘩𝘦𝘮𝘪𝘯𝘧𝘰𝘳𝘮𝘢𝘵𝘪𝘤𝘴 & 𝘴𝘮𝘢𝘭𝘭

Have we solved de novo peptide design? 1/🧵 RFpeptides designs macrocyclic peptide binders de novo against any protein target. This method follows from the binder design workflows of RFdiffusion, modifying it for peptides. X-ray structures of three complexes within 1.5 Å Cα

We are hiring! I don't tend to talk about the success of @tamarindbio publicly, but we are experiencing incredible demand. Tens of thousands of scientists are regularly using the platform, and we're onboarding GPUs and people as fast as we can. We just posted multiple roles

Effortless, rapid molecular dynamics and design with Tamarind AI Copilot Try it out now: app(.)tamarind(.)bio/copilot Molecular design and simulation traditionally requires extensive input preparation: Scientists have to write code to add ions, waters, membranes, parameterize

Should we stop humanizing antibodies? Recombinant DNA, and mAb production started the antibody engineering revolution, followed by chimerics and humanization via grafting antigen-binding regions to human frameworks. 1/🧵 Two decades apart, Clark (2000) and Getts et al.

A warning that proteins are highly dynamic molecules 1/🧵 Unlike what we get from AlphaFold, proteins don't exist as single, static molecules in the body. Structure prediction methods, while highly valuable, provide only a single moment in a protein's existence. There are

How Can We Computationally Solubilize Membrane Proteins? Thread 👇 Membrane proteins are not members of the soluble proteome, i.e. their structure cannot be maintained outside of the membrane/in solution. Traditionally, the solubilization would be done using detergent, but this

How Genentech/Prescient Design use AI to design 3-100x better antibodies Genentech uses lab in a loop as an iterative approach for designing antibodies by marrying generative machine learning models, property predictors and data from in vitro experimentation. Thread👇 The

When are AI-designed drugs making it to patients? No AI-designed medicine is on pharmacy shelves yet, but the first wave of molecules now in Phase 2/early Phase 3 (chiefly rentosertib for IPF). Here's where we are, a thread👇