AmyloDeep: protein language model-based ensemble model for predicting amyloid propensity from the amino acid…

Abstract

Motivation: Amyloids are predominantly β-sheet-rich, stable protein structures that can maintain their presence in the human body for multiple years. Amyloid protein aggregates contribute to the development of multiple neurodegenerative diseases, such as Alzheimer’s, Parkinson’s, and Huntington’s, and are involved in different vital functions, such as memory formation and immune system function.

Results: Here, we used advanced machine learning and deep learning techniques to predict amyloid propensity from the amino acid sequence. First, we aggregated labeled amino acid sequence data from multiple sources, obtaining a roughly balanced dataset of 2365 sequences for binary classification. We leveraged that data to both fine-tune the ESM2 model and to train new models based on protein embeddings from ESM2 and UniRep. The predictions from these models were then unified into a single soft voting ensemble model, yielding highly robust and accurate results. We further made a tool where users can provide the amino acid sequence and get the amyloid formation probabilities of different segments of the input sequence.

Availability and implementation: Users can access the light version of AmyloDeep through the web server at https://amylodeep.com/, and the full model is available as a Python package at https://pypi.org/project/amylodeep/.

Comparative Conformational Sampling of Apo Adenylate Kinase by Enhanced Molecular Dynamics and a Sequence-Based Generative Ensemble Model
Qingling Zhao, Ting Cui, Ying Li
Molecules1 Oct 2026Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationCrossrefEurope PMC
Predicting protein structures at any folding stage from sequence
Lisheng Zhang, Qingsong Li, Zinan Xie +7
arXiv3 Nov 2026PreprintCC BY 4.0Full textBiophysicsarXiv
An RNA Language Model trained on sequence alone reveals the structural logic of Internal Ribosome Entry Sites
A. Sychla, P. Bongrand, G. Yang +5
bioRxiv19 May 2026PreprintCC BYFull text1 citedMolecular biology & gene regulationbioRxiv
AIPFusion-Net: an interaction-fusion framework for anti-inflammatory peptide prediction using protein language model embeddings, local sequence patterns, and handcrafted descriptors.
Wentao Wang, Weidong Yang, Hongmei Li +2
Frontiers in molecular biosciences5 Oct 2026Peer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
Predicting protein electrostatics with protein language models
Mingzhe Shen, Guy W. Dayhoff II, Daniel Kortzak +1
Nature Communications24 Oct 2026Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationCrossref

Cite

BibTeX RIS

BibTeX

@article{davtyan2026amylodeep,
  title = {{AmyloDeep: protein language model-based ensemble model for predicting amyloid propensity from the amino acid sequence}},
  author = {Alisa Davtyan and Anahit Khachatryan and Rafayel Petrosyan},
  journal = {Bioinformatics Advances},
  year = {2026},
  month = oct,
  doi = {10.1093/bioadv/vbag296},
  url = {https://doi.org/10.1093/bioadv/vbag296},
  note = {CC BY 4.0},
}

RIS

TY  - JOUR
TI  - AmyloDeep: protein language model-based ensemble model for predicting amyloid propensity from the amino acid sequence
AU  - Alisa Davtyan
AU  - Anahit Khachatryan
AU  - Rafayel Petrosyan
PY  - 2026
DA  - 2026/10/06
JO  - Bioinformatics Advances
DO  - 10.1093/bioadv/vbag296
UR  - https://doi.org/10.1093/bioadv/vbag296
AB  - Motivation: Amyloids are predominantly β-sheet-rich, stable protein structures that can maintain their presence in the human body for multiple years. Amyloid protein aggregates contribute to the development of multiple neurodegenerative diseases, such as Alzheimer’s, Parkinson’s, and Huntington’s, and are involved in different vital functions, such as memory formation and immune system function.  Results: Here, we used advanced machine learning and deep learning techniques to predict amyloid propensity from the amino acid sequence. First, we aggregated labeled amino acid sequence data from multiple sources, obtaining a roughly balanced dataset of 2365 sequences for binary classification. We leveraged that data to both fine-tune the ESM2 model and to train new models based on protein embeddings from ESM2 and UniRep. The predictions from these models were then unified into a single soft voting ensemble model, yielding highly robust and accurate results. We further made a tool where users can provide the amino acid sequence and get the amyloid formation probabilities of different segments of the input sequence.  Availability and implementation: Users can access the light version of AmyloDeep through the web server at https://amylodeep.com/, and the full model is available as a Python package at https://pypi.org/project/amylodeep/.
SN  - 2635-0041
ER  -

Abstract

Related

Cite