ArticleNature methods2025
InterPLM: discovering interpretable features in protein language models via sparse autoencoders.
Article in Nature methods, 2025. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Cited by 38 papers.
What it found
Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.
The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.
The trial behind it
Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.
Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.
Who cites it
38 citing papers in PubMed.
- Mechanistic interpretability of fine-tuned protein language models for nanobody thermostability prediction.Bioinformatics (Oxford, England) · 2026Article
- Mechanistic Interpretability of Protein Language Models Reveals Encoded Structural and Functional Properties of Intrinsically Disordered Proteins.bioRxiv : the preprint server for biology · 2026Article
- EmmaEmb: A quantitative framework for analyzing embedding spaces in molecular biology.Patterns (New York, N.Y.) · 2026Article
- Unsupervised discovery of functional sequence patterns from protein language model with MotifAE.Nature communications · 2026Article
- From conservation scores to language models: a feature engineering perspective on machine learning for genetic variant interpretation.Briefings in bioinformatics · 2026Review
- Toward generalizable and interpretable AI in regulatory genomics.Nature genetics · 2026Review
- AlphaFold reveals but sometimes distorts an organizational principle of protein folding.bioRxiv : the preprint server for biology · 2026Article
- Scanning the horizon: deep mutational scanning approaches in virology.Journal of virology · 2026Review
- Generative AI for controllable protein sequence design: A survey.npj drug discovery · 2026Review
- PUFFIN: protein unit discovery with functional supervision.Bioinformatics (Oxford, England) · 2026Article
- Compositional and interpretable representation of histology using AI foundation models and sparse autoencoders.bioRxiv : the preprint server for biology · 2026Article
- amyloid-predict and LLPS-predict: Predicting phase separation propensities in the intrinsically disordered proteome.Proceedings of the National Academy of Sciences of the United States of America · 2026Article
- INBAdvanced science (Weinheim, Baden-Wurttemberg, Germany) · 2026Article
- AutomataGPT: Transformer-Based Forecasting and Ruleset Inference for Two-Dimensional Cellular Automata.Advanced science (Weinheim, Baden-Wurttemberg, Germany) · 2026Article
- AlphaInterp: Mechanistic Interpretability of AlphaFold 3 Reveals How Evolutionary Information Shapes Protein Structure Prediction.bioRxiv : the preprint server for biology · 2026Article
- Detecting and quantifying overparametrization in RNA language models with REDIAL.bioRxiv : the preprint server for biology · 2026Article
- Rapid directed evolution guided by protein language models and epistatic interactions.Science (New York, N.Y.) · 2026Article
- Generative design of intrinsically disordered protein regions with IDiom.bioRxiv : the preprint server for biology · 2026Article
- Inferring context-specific site variation with evotuned protein language models.NAR genomics and bioinformatics · 2026Article
- Deconvolving mutation effects on protein stability and function with disentangled protein language models.bioRxiv : the preprint server for biology · 2026Article
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
2 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
Despite their success in protein modeling and design, the internal mechanisms of protein language models (PLMs) are poorly understood. Here we present a systematic framework to extract and analyze interpretable features from PLMs using sparse autoencoders. Training sparse autoencoders on ESM-2 embeddings, we identify thousands of interpretable features highlighting biological concepts including binding sites, structural motifs and functional domains. Individual neurons show considerably less conceptual alignment, suggesting PLMs store concepts in superposition. This superposition persists across model scales and larger PLMs capture more interpretable concepts. Beyond known annotations, ESM-2 learns coherent patterns across evolutionarily distinct protein families. To systematically analyze these numerous features, we developed an automated interpretation approach using large language models for feature description and validation. As practical applications, these features can accurately identify missing database annotations and enable targeted steering of sequence generation. Our results show PLM representations can be decomposed into interpretable components, demonstrating the feasibility and utility of mechanistically interpreting these models.
Indexed as
Identifiers
41023434What OpenQuestion holds
Registered trials
Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.