Ensemble-AHTPpred: A Robust Ensemble Machine Learning Model Integrated With a New Composite Feature for Identifying...

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Hypertension is a global health crisis, increasing risks for cardiovascular and other diseases.
• Antihypertensive peptides (AHTPs) from natural sources offer a promising, low-side-effect approach for prevention and treatment.
• Experimental identification of AHTPs is laborious; computational methods are needed for efficient screening.

Purpose of the Study:

• To develop a robust computational tool, Ensemble-AHTPpred, for predicting antihypertensive peptides (AHTPs).
• To integrate multiple machine learning algorithms for enhanced prediction accuracy and reliability.
• To utilize a comprehensive feature set for improved characterization of AHTPs.

Main Methods:

• An ensemble machine learning model combining Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGB).
• Inclusion of diverse computed features: physicochemical properties, amino acid compositions (AACs), transitions, n-grams, and secondary structure information.
• Integration of a novel composite feature derived from a logistic regression function for enhanced peptide characterization.

Main Results:

• Ensemble-AHTPpred achieved over 90% accuracy on independent test datasets.
• The tool demonstrated high precision in predicting novel, experimentally validated AHTPs not included in training or testing.
• The ensemble approach enhanced the robustness and predictive power compared to individual algorithms.

Conclusions:

• Ensemble-AHTPpred is a highly accurate and efficient tool for identifying potential antihypertensive peptides.
• This computational approach can accelerate the discovery of functional AHTPs from natural sources.
• The developed method aids in the search for novel nutraceuticals for hypertension management.