Rore: robust and efficient antioxidant protein classification via a novel dimensionality reduction strategy based on...

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning in Proteomics

Background:

• Protein identification relies on efficient classification with minimal features.
• Traditional feature selection methods often lead to information loss, hindering classifier performance.
• Dimensionality reduction is crucial for enhancing classification efficiency in proteomics.

Purpose of the Study:

• To introduce Rore, a novel algorithm for feature-dimensionality reduction in protein identification.
• To overcome the information loss issue inherent in conventional feature selection techniques.
• To improve the accuracy and efficiency of protein classification models.

Main Methods:

• Rore employs a feature-dimensionality reduction strategy by mapping original features to a latent space.
• This method retains essential feature information while reducing the number of representations.
• The algorithm was validated using an antioxidant protein dataset.

Main Results:

• Rore achieved high performance on an antioxidant protein dataset, with 95.88% accuracy.
• The algorithm demonstrated a Matthew's Correlation Coefficient (MCC) of 91.78%.
• Excellent results were obtained using vectors with only 15 features, showcasing efficiency.

Conclusions:

• Rore effectively preserves original feature information, mitigating information loss.
• The algorithm offers a significant improvement over traditional feature selection methods in protein identification.
• Rore provides a powerful and efficient tool for protein classification, available online.