Related Experiment Video
Updated: Feb 28, 2026
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
BiToxNet: a deep learning framework integrating multimodal features for accurate identification of neurotoxic
Feng Wang1,2, Peilin Xie3,4, Xingqiao Lin2
1School of Informatics, Xiamen University, 361005, Xiamen, China.
Background:
Accurate prediction of the neurotoxicity of peptides and proteins is critically important for the safety assessment of protein therapeutics and the development of protein-based drugs. Although experimental methods can reliably identify neurotoxic peptides and neurotoxins, they are labor-intensive, costly, and unsuitable for large-scale screening. Existing computational approaches are often limited by shallow feature engineering and suboptimal multimodal fusion strategies, which restrict their predictive accuracy and generalizability in real-world applications.
Results:
In this study, we propose BiToxNet, a deep learning framework that integrates evolutionary embeddings derived from a protein large language model with ten handcrafted biochemical descriptors through a bilinear attention network (BAN). This design enables effective modeling of cross-modal interactions and residue-level dependencies critical for neurotoxicity prediction. BiToxNet was evaluated on three datasets of different sequence lengths, namely Protein, Peptide, and Combined datasets, achieving accuracies of 92.3%, 96.0%, and 92.7%, respectively, and consistently outperforming existing state-of-the-art methods. Ablation studies confirmed the importance of both evolutionary embeddings and handcrafted features, as well as the critical role of BAN in feature fusion. Visualization analyses using t-SNE and hierarchical clustering further demonstrated that BiToxNet learns highly discriminative representations without reliance on domain-specific prior knowledge. Additional evaluation on an external imbalanced dataset validated the robustness and strong generalization capability of the proposed framework.
Conclusions:
Overall, BiToxNet provides a powerful and generalizable computational framework for the accurate identification of neurotoxic peptides and proteins. By effectively integrating evolutionary and biochemical information through bilinear attention, BiToxNet offers a valuable tool for neurotoxin screening and protein drug safety assessment, and presents a distinctive modeling strategy applicable to a wide range of biological sequence analysis tasks.
Insights
BiToxNet, a novel deep learning framework, accurately predicts peptide and protein neurotoxicity by integrating evolutionary and biochemical data. This computational tool enhances safety assessments for protein therapeutics and aids in drug development.
Area of Science:
- Computational biology
- Bioinformatics
- Machine learning in drug discovery
Background:
- Accurate prediction of peptide and protein neurotoxicity is crucial for drug safety and development.
- Experimental methods are costly and time-consuming for large-scale screening.
- Existing computational methods lack accuracy due to limited feature engineering and fusion strategies.
Purpose of the Study:
- To develop a robust deep learning framework for predicting neurotoxicity.
- To improve the accuracy and generalizability of computational neurotoxicity prediction.
- To provide a valuable tool for neurotoxin screening and protein drug safety.
Main Methods:
- Developed BiToxNet, a deep learning framework integrating evolutionary embeddings from a protein large language model and ten handcrafted biochemical descriptors.
- Employed a bilinear attention network (BAN) to model cross-modal interactions and residue-level dependencies.
- Evaluated BiToxNet on Protein, Peptide, and Combined datasets of varying sequence lengths.
Main Results:
- BiToxNet achieved high accuracies: 92.3% (Protein), 96.0% (Peptide), and 92.7% (Combined).
- The framework consistently outperformed existing state-of-the-art methods.
- Ablation studies and visualization analyses confirmed the importance of integrated features and BAN, demonstrating robust generalization capabilities.
Conclusions:
- BiToxNet offers a powerful and generalizable computational framework for identifying neurotoxic peptides and proteins.
- The integration of evolutionary and biochemical information via bilinear attention provides a novel modeling strategy.
- BiToxNet serves as a valuable tool for neurotoxin screening and protein therapeutic safety assessment.
Related Concept Videos
01:33Peptide Identification Using Tandem Mass Spectrometry
This technique helps gather information regarding the protein from which the peptide was obtained and to study the peptides’ amino acid sequence. Identifying peptides from a complex mixture is an important component of the growing field of...
01:30Classification of Neurotransmitters