Sequence homology score-based deep fuzzy network for identifying therapeutic peptides

Area of Science:

• Biomedical Science
• Computational Biology
• Bioinformatics

Background:

• Therapeutic peptide detection is crucial but conventional methods are slow.
• Computational biology offers efficiency improvements for peptide detection.
• Existing computational methods often neglect noise, impacting generalization.

Purpose of the Study:

• To develop an advanced computational model for enhanced therapeutic peptide detection.
• To improve the generalization performance of therapeutic peptide prediction.
• To introduce a novel sequence homology score-based deep fuzzy echo-state network with maximizing mixture correntropy (SHS-DFESN-MMC) model.

Main Methods:

• Developed a SHS-DFESN-MMC model incorporating sequence homology and fuzzy echo-state networks.
• Utilized maximizing mixture correntropy to mitigate noise effects.
• Validated the model on eight diverse therapeutic peptide datasets using 10-fold cross-validation and independent test sets.

Main Results:

• The SHS-DFESN-MMC model achieved the highest average area under the receiver operating characteristic curve (AUC) values.
• Achieved an average AUC of 0.926 on training sets and 0.923 on independent test sets across all datasets.
• Demonstrated superior performance compared to existing computational methods for therapeutic peptide prediction.

Conclusions:

• The SHS-DFESN-MMC model significantly enhances therapeutic peptide detection accuracy and generalization.
• This computational approach offers a more efficient and robust alternative to traditional experimental methods.
• The findings highlight the potential of integrating deep learning with noise-robust techniques for biomedical applications.