Mito-GSAAC: mitochondria prediction using genetic ensemble classifier and split amino acid composition.

Area of Science:

• Bioinformatics
• Molecular Cell Biology
• Proteomics

Background:

• Mitochondria are vital organelles implicated in cellular mortality and various human diseases.
• Accurate identification of mitochondrial proteins is essential for biological research and drug development.
• Existing techniques for mitochondrial protein prediction require enhancement in accuracy and reliability.

Purpose of the Study:

• To develop an effective feature extraction strategy for mitochondrial protein prediction.
• To create an ensemble approach that leverages advanced feature extraction for improved mitochondria classification.
• To propose the Mito-GSAAC system for enhanced prediction of mitochondrial proteins.

Main Methods:

• Investigated four protein representations: amino acid composition, dipeptide composition, pseudo amino acid composition, and split amino acid composition (SAAC).
• Trained individual classifiers including SVM, k-nearest neighbor, multilayer perceptron, random forest, AdaBoost, and bagging.
• Developed an ensemble classifier using genetic programming (GP) to combine individual classifier decision spaces.

Main Results:

• The GP-based ensemble classifier using SAAC features achieved a prediction accuracy of 92.62% on the Mitochondria dataset.
• On the Malaria Parasite Mitochondria dataset, SVM with SAAC achieved 93.21% accuracy, further enhanced by the GP-based ensemble.
• Split amino acid composition (SAAC) demonstrated superior discrimination power compared to other feature extraction strategies for mitochondria prediction.

Conclusions:

• The Mito-GSAAC system, integrating SAAC features and GP-based ensemble, significantly improves mitochondrial protein prediction accuracy.
• SAAC's effectiveness in discriminating mitochondrial from non-mitochondrial proteins, particularly at termini, contributes to enhanced prediction.
• The developed predictor is expected to impact fields such as Molecular Cell Biology, Proteomics, Bioinformatics, System Biology, and Drug Development.