Related Experiment Video
Updated: Sep 10, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.9K
An innovative peptide toxicity prediction model based on multi-scale convolutional neural network and residual
Shengli Zhang1, Jingyi Ren1, Yunyun Liang2
1School of Mathematics and Statistics, Xidian University, Xi'an 710071, P. R. China.
Bioinformatics (Oxford, England)
|August 22, 2025
Summary
A new model, ToxMSRC, accurately predicts peptide toxicity using advanced AI techniques. This tool aids in developing safer peptide therapeutics by identifying potential toxic sequences early in the drug development process.
Area of Science:
- Computational biology
- Bioinformatics
- Drug discovery
Background:
- Peptide toxicity poses significant risks in therapeutic development, potentially causing organ damage and immune reactions.
- Accurate prediction of peptide toxicity is crucial for ensuring drug safety and efficacy.
Purpose of the Study:
- To develop a novel computational model for predicting peptide toxicity.
- To improve the accuracy and reliability of peptide toxicity assessments in drug development.
Main Methods:
- Utilized a hybrid approach combining continuous bag of words (CBOW), synthetic minority over-sampling technique (SMOTE), multi-scale convolutional neural networks (CNN), and bidirectional long short-term memory (BiLSTM).
- Implemented a residual connection to enhance model generalization and prevent overfitting.
- Addressed data imbalance issues by augmenting positive samples.
Main Results:
- The ToxMSRC model achieved high predictive performance, with BACC scores of 92.17% on independent test1 and 86.89% on independent test2.
- Outperformed existing state-of-the-art models in peptide toxicity prediction.
- Provided insights into the sequence-toxicity relationship.
Conclusions:
- ToxMSRC offers a robust and accurate method for predicting peptide toxicity.
- The model demonstrates significant potential for practical application in peptide-based drug development.
- Open-source code and data are available for community use and further research.
