💡 Research

Protein complexes are central to cellular function, but experimental determination of their structures remains challenging. Structure prediction methods require prior knowledge of stoichiometry - the number of copies of each protein entity within a complex. Current approaches rely on computationally expensive brute-force methods that run structure prediction on multiple stoichiometry combinations, often with limited accuracy. We introduce Stoic, a method that uses protein language model embeddings to predict protein complex stoichiometry. Our approach learns to identify interface residues that participate in protein-protein interactions, rather than relying on global sequence features. By integrating these interface-aware embeddings into a graph neural network, Stoic achieves fast and accurate stoichiometry prediction for both homomeric and heteromeric targets.

Daniil Litvinov, Lorenzo Pantolini, Peter Škrinjar, Gerardo Tauriello, Caitlyn L. McCafferty, Benjamin D. Engel, Torsten Schwede, Janani Durairaj

Deep learning has driven major breakthroughs in protein structure prediction, however the next critical advance is accurately predicting how proteins interact with other molecules, especially small molecule ligands, to enable real-world applications such as drug discovery and design. Recent deep learning all-atom methods have been built to address this challenge, but evaluating their performance on the prediction of protein-ligand complexes has been inconclusive due to the lack of relevant benchmarking datasets. Here we present a comprehensive evaluation of four leading all-atom cofolding deep learning methods using our newly introduced benchmark dataset Runs N' Poses, which comprises 2,600 high-resolution protein-ligand systems released after the training cutoff used by these methods. We demonstrate that current co-folding approaches largely memorise ligand poses from their training data, hindering their use for de novo drug design.

Peter Škrinjar, Jérôme Eberhardt, Gabriel Studer, Gerardo Tauriello, Torsten Schwede, Janani Durairaj

Protein-ligand interactions (PLI) are foundational to small molecule drug design. With computational methods striving towards experimental accuracy, there is a critical demand for a well-curated and diverse PLI dataset. Existing datasets are often limited in size and diversity, and commonly used evaluation sets suffer from training information leakage, hindering the realistic assessment of method generalization capabilities. To address these shortcomings, we present PLINDER, the largest and most annotated dataset to date, comprising 449,383 PLI systems, each with over 500 annotations, similarity metrics at protein, pocket, interaction and ligand levels, and paired unbound (apo) and predicted structures. We propose an approach to generate training and evaluation splits that minimizes task-specific leakage and maximizes test set quality, and compare the resulting performance of DiffDock when retrained with different kinds of splits.

Janani Durairaj, Yusuf Adeshina, Zhonglin Cao, Xuejin Zhang, Vladas Oleinikovas, Thomas Duignan, Zachary McClure, Xavier Robin, Gabriel Studer, Daniel Kovtun, Emanuele Rossi, Guoqing Zhou, Srimukh Veccham, Clemens Isert, Yuxing Peng, Prabindh Sundareson, Mehmet Akdel, Gabriele Corso, Hannes Stärk, Gerardo Tauriello, Zachary Carpenter, Michael Bronstein, Emine Kucukbenli, Torsten Schwede, Luca Naef

Detecting remote homology with speed and sensitivity is crucial for tasks like function annotation and structure prediction. We introduce a novel approach using contrastive learning to convert protein language model embeddings into a new 20-letter alphabet, TEA, enabling highly efficient large-scale protein homology searches. Searching with our alphabet performs on par with and complements structure-based methods without requiring any structural information, and with the speed of sequence search. Ultimately, we bring the exciting advances in protein language model representation learning to the plethora of sequence bioinformatics algorithms developed over the past century, offering a powerful new tool for biological discovery.

Lorenzo Pantolini, Gabriel Studer, Laura Engist, Ieva Pudziuvelytė, Florian Pommerening, Andrew Mark Waterhouse, Gerardo Tauriello, Martin Steinegger, Torsten Schwede, Janani Durairaj

We are now entering a new era in protein sequence and structure annotation, with hundreds of millions of predicted protein structures made available through the AlphaFold database 1 . These models cover nearly all proteins that are known, including those challenging to annotate for function or putative biological role using standard homology-based approaches. In this study, we examine the extent to which the AlphaFold database has structurally illuminated this ‘dark matter’ of the natural protein universe at high predicted accuracy. We further describe the protein diversity that these models cover as an annotated interactive sequence similarity network, accessible at https://uniprot3d.org/atlas/AFDB90v4 . By searching for novelties from sequence, structure and semantic perspectives, we uncovered the β-flower fold, added several protein families to Pfam database 2 and experimentally demonstrated that one of these belongs to a new superfamily of translation-targeting toxin–antitoxin systems, TumE–TumA. This work underscores the value of large-scale efforts in identifying, annotating and prioritizing new protein families. By leveraging the recent deep learning revolution in protein bioinformatics, we can now shed light into uncharted areas of the protein universe at an unprecedented scale, paving the way to innovations in life sciences and biotechnology.

Protein scaffolds around fixed functional motifs, optionally designed in complex with a target. MotifCraft combines AlphaFold2-guided optimization with ProteinMPNN sequence design and structure validation.

Océane Follonier, Torsten Schwede, Janani Durairaj

De novo protein design expands the functional protein universe beyond natural evolution, offering vast therapeutic and industrial potential. Monte Carlo sampling in protein design is under-explored due to the typically long simulation times required or prohibitive time requirements of current structure prediction oracles. Here we make use of a 20-letter structure-inspired alphabet derived from protein language model embeddings to score random mutagenesis-based Metropolis sampling of amino acid sequences. This facilitates fast template-guided and unconditional design, generating sequences that satisfy in silico designability criteria without known homologues. Ultimately, this unlocks a new path to fast and de novo protein design.

Lorenzo Pantolini, Janani Durairaj