EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

PGAT-ABPp: harnessing protein language models and graph attention networks for antibacterial peptide identification with remarkable accuracy

Yuelei Hao iD, Xuyang Liu, Haohao Fu, Xueguang Shao, Wensheng Cai iD

Bioinformatics · 2024

DOI: 10.1093/bioinformatics/btae497 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioinformatics, 2024. Verfasst von Yuelei Hao und 4 weiteren Autor:innen. Beteiligte Einrichtung: Research Center for Analytical Sciences, Tianjin Key Laboratory of Biosensing and Molecular Recognition, State Key Laboratory of Medicinal Chemical B…. Hauptthemen laut MEDLINE: Computational Biology. Die Arbeit wurde bisher 33-mal zitiert (Crossref) und nennt 66 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Haihe Laboratory of Sustainable Chemical Transformations.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Motivation The emergence of drug-resistant pathogens represents a formidable challenge to global health. Using computational methods to identify the antibacterial peptides (ABPs), an alternative antimicrobial agent, has demonstrated advantages in further drug design studies. Most of the current approaches, however, rely on handcrafted features and underutilize structural information, which may affect prediction performance. Results To present an ultra-accurate model for ABP identification, we propose a novel deep learning approach, PGAT-ABPp. PGAT-ABPp leverages structures predicted by AlphaFold2 and a pretrained protein language model, ProtT5-XL-U50 (ProtT5), to construct graphs. Then the graph attention network (GAT) is adopted to learn global discriminative features from the graphs. PGAT-ABPp outperforms the other fourteen state-of-the-art models in terms of accuracy, F1-score and Matthews Correlation Coefficient on the independent test dataset. The results show that ProtT5 has significant advantages in the identification of ABPs and the introduction of spatial information further improves the prediction performance of the model. The interpretability analysis of key residues in known active ABPs further underscores the superiority of PGAT-ABPp. Availability and implementation The datasets and source codes for the PGAT-ABPp model are available at https://github.com/moonseter/PGAT-ABPp/.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Computational BiologyAntimicrobial PeptidesAnti-Bacterial AgentsDeep Learning

Substanzen

Antimicrobial PeptidesAnti-Bacterial Agents

Autor:innen und Institutionen

  • Research Center for Analytical Sciences, Tianjin Key Laboratory of Biosensing and Molecular Recognition, State Key Laboratory of Medicinal Chemical Biology, College of Chemistry, Nankai University, Tianjin 300071, China.
  • Haihe Laboratory of Sustainable Chemical Transformations, Tianjin 300192, China.

ORCID-Profile: Yuelei Hao, Wensheng Cai

Förderung

  • National Natural Science Foundation of China
  • Haihe Laboratory of Sustainable Chemical Transformations

Zitieren

APA

Hao, Y., Liu, X., Fu, H., Shao, X., & Cai, W. (2024). PGAT-ABPp: harnessing protein language models and graph attention networks for antibacterial peptide identification with remarkable accuracy. Bioinformatics. https://doi.org/10.1093/bioinformatics/btae497

Vancouver

Hao Y, Liu X, Fu H, Shao X, Cai W. PGAT-ABPp: harnessing protein language models and graph attention networks for antibacterial peptide identification with remarkable accuracy. Bioinformatics. 2024. doi:10.1093/bioinformatics/btae497

BibTeX
@article{Hao2024,
  author = {Hao, Yuelei and Liu, Xuyang and Fu, Haohao and Shao, Xueguang and Cai, Wensheng},
  title = {PGAT-ABPp: harnessing protein language models and graph attention networks for antibacterial peptide identification with remarkable accuracy},
  journal = {Bioinformatics},
  year = {2024},
  doi = {10.1093/bioinformatics/btae497}
}
Kostenlos starten →