From scarce functional labels to label-aware generation in homologous protein families
Open Access
- 1 October 2026
- journal article
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 7 (5) , 055050
Abstract
Accurately annotating and controlling protein function from sequence data remains a major challenge in protein engineering, especially when functional labels are scarce within large homologous families. Here, we study a two-stage light-supervision strategy for fine-grained functional annotation and label-aware sequence generation. First, we compare several sequence representations, including one-hot encodings, restricted Boltzmann machines (RBMs) and protein language model (pLM) embeddings, for predicting intra-family specificity labels from limited supervision. By using train/test splits that explicitly reduce phylogenetic leakage, we show that pLM-based representations do not systematically outperform family-specific RBM embeddings or even simple one-hot baselines in this regime. Second, we use the inferred annotations to train an annotation-aware RBM capable of generating artificial homologs conditioned on prescribed labels. For a specific protein family, we quantify how the number and quality of available labels determine the reliability of conditional generation. Our results show that scarce annotations can support label-aware protein design when they are accurately propagated, while also highlighting the importance of phylogeny-aware evaluation for assessing functional annotation methods within homologous families.Keywords
Funding Information
- Horizon Europe MSCA Staff Exchange (101131463)
This publication has 31 references indexed in Scilit:
- Deep-learning-based design of synthetic orthologs of SH3 signaling domainsCell Systems, 2024
- Regularization, early-stopping and dreaming: A Hopfield-like setup to address generalization and overfittingNeural Networks, 2024
- Evaluating large language models for annotating proteinsBriefings in Bioinformatics, 2024
- Computational Approaches to Predict Protein–Protein Interactions in Crowded Cellular EnvironmentsChemical Reviews, 2024
- Latent generative landscapes as maps of functional diversity in protein sequence spaceNature Communications, 2023
- Efficient generative modeling of protein sequences using simple autoregressive modelsNature Communications, 2021
- Highly accurate protein structure prediction with AlphaFoldNature, 2021
- Evolutionary Models of Fold-Switching ProteinsBiophysical Journal, 2021
- Exploring protein fitness landscapes by directed evolutionNature Reviews Molecular Cell Biology, 2009
- Crystal structure of a Src-homology 3 (SH3) domainNature, 1992