Burkhard Rost

The paper investigates whether functional frustration in proteins is an evolutionary adaptation or a byproduct of fold architecture, finding that frustration hotspots are inherent physical constraints co-opted for function.

A fine-tuned ProtT5 protein language model predicts per-residue energetic frustration from sequence (Pearson r 0.86 on unseen CATH topologies) and profiles the human proteome in ~17 minutes on one GPU.

DeePEn evaluates protein engineering models' ability to predict fitness of variants with increasing mutational distance from wildtype.

This work analyzes CAFA5 submissions, showing improved accuracy in protein function prediction and highlights the role of protein language models and 3D structures in performance gains.

This study uses pseudoperplexity to assess whether ProtT5 memorizes its training data, finding detectable but limited memorization.

ProtSpace visualizes protein language model embeddings interactively, enabling hypothesis generation across biological scales without installation.

A ProtT5-embedding, ConvNext-plus-attention ensemble predicts human pathogenic capacity from whole bacterial genomes, outperforming prior predictors on species absent from training and flagging influential proteins.

This study identifies three conserved genomic regions in ants as evolutionary hotspots for venom genes, revealing repeated recruitment to conserved platforms and diverse evolutionary dynamics.

Biocentral provides an open embedding-based service for protein predictions, demonstrating its utility with a virus database analysis and training module.

UdonPred predicts per-residue protein disorder from sequence using embeddings from ProstT5, showing that annotation quality, not model capacity, drives performance in continuous disorder prediction.

This paper introduces ProtProfileMD, a model that predicts protein dynamics from sequences, improving homology detection by capturing conformational changes through 3Di tokens.

ProFam is an open-source protein family language model that improves fitness prediction and design by leveraging evolutionary context through homologous sequences.

This study benchmarks fourteen protein-language models (pLMs) to determine which best captures biological properties, finding that mid-scale models often outperform larger ones for general tasks.

FlatProt provides 2D visualizations for comparing protein structures, enabling scalable and standardized analysis of large protein sets.

The paper evaluates missense variant effect predictors using CAGI6 data, finding strong performance across settings and highlighting biases in some predictors.

TMVisDB provides a database and visualization tool for transmembrane proteins, enabling large-scale analysis of predicted structures and topologies without requiring costly AlphaFold3-style predictions.

The paper analyzes the DNA sequence and methylome of Jonathan, a 192-year-old giant tortoise, to identify epigenomic features associated with extreme longevity.

ProtSpace visualizes protein embeddings in 2D/3D to reveal functional patterns and challenges existing classifications in phage and venom protein analyses.

VespaG is a fast missense variant effect predictor using protein Language Models, achieving high accuracy on the ProteinGym benchmark.

ProstT5, a ProtT5 fine-tuned to translate between amino acids and Foldseek 3Di tokens, predicts structure strings for remote homology search and performs inverse folding.

This study shows that fine-tuning protein language models improves prediction performance across tasks and can be done efficiently, especially for small datasets.

The paper introduces a machine learning model, IDBindT5, to predict binding residues in intrinsically disordered protein regions, achieving balanced accuracy comparable to state-of-the-art methods.

MutAmore generates 3D mutation movies of proteins by predicting structural conformations of all possible single amino-acid variants and visualizing them as color-coded animations.

The paper predicts mutational outcomes for 1.5M missense variants across 72 protein families using alignment-based methods, showing high-quality and compute-efficient results for entire proteomes.

The paper uses AlphaFold2 to analyze protein structures from 21 organisms, identifying 26 novel superfamilies and expanding structural diversity insights.

Protein-specific tuning of masking, depth and positional embeddings, rather than more parameters, yields an encoder-decoder model that tops structure and function benchmarks at a fraction of the compute.

The paper introduces CATHe, a method using protein language model embeddings to detect remote homologues for CATH superfamilies, achieving high accuracy and validating predictions with AlphaFold structures.

The paper reconstructs the evolutionary history of three-finger toxins (3FTxs) and identifies their non-toxin ancestor as a non-secretory Ly6 in squamate reptiles.

LambdaPP provides fast, accessible protein feature predictions including structure and function, using AI models and embeddings from protein Language Models.

TMvisDB provides a resource for browsing and visualizing predicted transmembrane proteins with topological annotations, using AlphaFold DB and TMbed predictions.