Beyond profiles: supervised repeat annotation using protein embeddings
- 1Max Planck Institute for Biology
Abstract
Protein repeats give rise to diverse protein structures and functions, yet their sequence-based detection is often challenging because individual repeat units can diverge substantially in sequence while retaining similar structures. Most sensitive approaches rely on sequence information directly represented by alignments or derived profiles, thereby limiting flexibility and scalability. Here, we introduce TREAD (Transfer learning-based REpeat Annotation using Protein EmbeDdings), a supervised framework that reformulates repeat detection as a residue-level annotation problem and operates directly on embeddings from protein language models. Instead of constructing explicit probabilistic profiles, TREAD learns repeat-specific features implicitly, enabling residue-resolved scoring and flexible repeat segment extraction. Across complementary benchmarks based on RepeatsDB and Pfam, TREAD consistently matches or outperforms the widely used profile-based tool HMMER, particularly in low-data and high-divergence settings. The model exhibits robustness to score thresholding and demonstrates better generalization to independent test sets, including those containing remotely related families. To illustrate the practical utility of TREAD, we applied it to survey {beta}-propeller proteins across the AlphaFold Database and representative proteomes to generate a comprehensive census of this fold. This analysis highlights extensive propeller diversity, identifies lineage-specific expansion patterns across the tree of life, and suggests previously unrecognized relationships between propellers and other repeat folds. Together, TREAD provides a flexible and scalable alternative to profile-based repeat annotation and establishes a general motif-centric framework for protein sequence annotation.
Related
- AMADEUS: Annotation-free multi-animal direction estimation using self-supervised learning
- SpacePath enables pathway and molecular-family analysis beyond annotation ambiguity in spatial metabolomics and lipidomics
- PalmLab: A Comprehensive Computational Platform for Systematic Annotation and Functional Interrogation of Protein Palmitoylation
- A Context-Aware 14-3-3 Binding Predictor Enabled by Data Augmentation, Multi-Scale Biological Features, and Protein Language Model Embeddings
- UHPLC-TIMS-qTOF-MS/MS expands metabolite annotation and identifies distinct cultivar-specific metabolite profiles in American elderberry flowers
BibTeX
@misc{qiu2026beyond,
title = {{Beyond profiles: supervised repeat annotation using protein embeddings}},
author = {K. Qiu and J. Ludwiczak and A. N. Lupas and S. Dunin-Horkawicz},
howpublished = {bioRxiv},
year = {2026},
month = may,
doi = {10.64898/2026.05.19.725729},
url = {https://www.biorxiv.org/content/10.64898/2026.05.19.725729v2},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Beyond profiles: supervised repeat annotation using protein embeddings
AU - K. Qiu
AU - J. Ludwiczak
AU - A. N. Lupas
AU - S. Dunin-Horkawicz
PY - 2026
DA - 2026/05/19
PB - bioRxiv
DO - 10.64898/2026.05.19.725729
UR - https://www.biorxiv.org/content/10.64898/2026.05.19.725729v2
AB - Protein repeats give rise to diverse protein structures and functions, yet their sequence-based detection is often challenging because individual repeat units can diverge substantially in sequence while retaining similar structures. Most sensitive approaches rely on sequence information directly represented by alignments or derived profiles, thereby limiting flexibility and scalability. Here, we introduce TREAD (Transfer learning-based REpeat Annotation using Protein EmbeDdings), a supervised framework that reformulates repeat detection as a residue-level annotation problem and operates directly on embeddings from protein language models. Instead of constructing explicit probabilistic profiles, TREAD learns repeat-specific features implicitly, enabling residue-resolved scoring and flexible repeat segment extraction. Across complementary benchmarks based on RepeatsDB and Pfam, TREAD consistently matches or outperforms the widely used profile-based tool HMMER, particularly in low-data and high-divergence settings. The model exhibits robustness to score thresholding and demonstrates better generalization to independent test sets, including those containing remotely related families. To illustrate the practical utility of TREAD, we applied it to survey {beta}-propeller proteins across the AlphaFold Database and representative proteomes to generate a comprehensive census of this fold. This analysis highlights extensive propeller diversity, identifies lineage-specific expansion patterns across the tree of life, and suggests previously unrecognized relationships between propellers and other repeat folds. Together, TREAD provides a flexible and scalable alternative to profile-based repeat annotation and establishes a general motif-centric framework for protein sequence annotation.
ER -