Omics-scale decoding of protein N-glycosylation using protein language model embeddings

Abstract

Protein language models (pLMs) learn rich sequence semantics from vast volume of protein sequences, yet the information encoded in their high-dimensional embeddings remains largely a black box, posing an interpretability challenge. Here we show that semantic features learned by pLMs can be decoded to reveal biologically meaningful signals using coevolutionary analysis without downstream supervised-training. Building on this insight, we developed MaskNGly, a zero-shot framework that identifies protein N-glycosylation sites directly from PLM-embeddings while achieving performance comparable to state-of-the-art supervised predictors. MaskNGly is computationally efficient and intrinsically transferable, enabling omics-scale characterization of protein glycosylation landscape. Applied across proteomes, it reveals that canonical N-X-S/T motifs gain higher glycosylation propensities with increasing organismal complexity and that glycosylation preferentially occurs at structured, solvent-exposed regions. Mutations altering high-scoring motifs are more frequently disease-associated. The framework also captures signatures of glycosylation-mediated protein allostery. These results establish an interpretable and scalable route to study glycobiology from pLM-embeddings.

A Context-Aware 14-3-3 Binding Predictor Enabled by Data Augmentation, Multi-Scale Biological Features, and Protein Language Model Embeddings
L. Liu, Z. Wang, X. Huang
bioRxiv18 Sep 2026PreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction
M. Marijan, I. Tanasijevic
bioRxiv31 Aug 2026PreprintCC BY-NCFull textBioinformatics & computational biologybioRxiv
CHORUS: A proteome-scale atlas of membrane interaction interfaces constructed using cross-protein language model embeddings from sequences alone
J. Zhao, S. Huang, J. Sun
bioRxiv4 Oct 2026PreprintCC BY-NC-NDFull textBioinformatics & computational biologybioRxiv
An intrinsically disordered proline-rich domain acts in synergy with AGP-type glycosylation to control the protein fate of FLA4
C. Hiremath, H. Xue, T. Acet +1
bioRxiv6 Oct 2026PreprintAll rights reservedFull textPlant biologybioRxiv
Deciphering the mechanisms of macrophage polarization through comprehensive analysis of cellular and cell-surface protein glycosylation.
Xing Xu, Yue Wu, Kejun Yin +1
Nature communications14 Sep 2026Peer-reviewedCC BY-NC-NDFull textImmunologyEurope PMC

Cite

BibTeX RIS

BibTeX

@misc{zhou2026omicsscale,
  title = {{Omics-scale decoding of protein N-glycosylation using protein language model embeddings}},
  author = {X. Zhou and X. Guan and Y. Zhou and J. Wang and Z. Peng and W. Ren and W. Wang and R. Xie and W. Li},
  howpublished = {bioRxiv},
  year = {2026},
  month = nov,
  doi = {10.64898/2026.11.02.763593},
  url = {https://www.biorxiv.org/content/10.64898/2026.11.02.763593v1},
  note = {CC BY-NC-ND},
}

RIS

TY  - UNPB
TI  - Omics-scale decoding of protein N-glycosylation using protein language model embeddings
AU  - X. Zhou
AU  - X. Guan
AU  - Y. Zhou
AU  - J. Wang
AU  - Z. Peng
AU  - W. Ren
AU  - W. Wang
AU  - R. Xie
AU  - W. Li
PY  - 2026
DA  - 2026/11/02
PB  - bioRxiv
DO  - 10.64898/2026.11.02.763593
UR  - https://www.biorxiv.org/content/10.64898/2026.11.02.763593v1
AB  - Protein language models (pLMs) learn rich sequence semantics from vast volume of protein sequences, yet the information encoded in their high-dimensional embeddings remains largely a black box, posing an interpretability challenge. Here we show that semantic features learned by pLMs can be decoded to reveal biologically meaningful signals using coevolutionary analysis without downstream supervised-training. Building on this insight, we developed MaskNGly, a zero-shot framework that identifies protein N-glycosylation sites directly from PLM-embeddings while achieving performance comparable to state-of-the-art supervised predictors. MaskNGly is computationally efficient and intrinsically transferable, enabling omics-scale characterization of protein glycosylation landscape. Applied across proteomes, it reveals that canonical N-X-S/T motifs gain higher glycosylation propensities with increasing organismal complexity and that glycosylation preferentially occurs at structured, solvent-exposed regions. Mutations altering high-scoring motifs are more frequently disease-associated. The framework also captures signatures of glycosylation-mediated protein allostery. These results establish an interpretable and scalable route to study glycobiology from pLM-embeddings.
ER  -

Abstract

Related

Cite