ApoPred

ApoPred predicts apolipoproteins and their subfamilies from protein sequences to support research into apolipoprotein roles in hyperlipidemia, atherosclerosis, Alzheimer's disease, and diabetes.

Key Features:

• Benchmark dataset: Uses a curated dataset of 270 known apolipoproteins and 535 non-apolipoproteins for model training and evaluation.
• Feature extraction: Represents sequences using pseudo-amino acid composition (PseAAC) and composition of k-spaced amino acid pairs (CKSAAP).
• Feature optimization: Ranks features with analysis of variance (ANOVA) and applies incremental feature selection to identify optimal subsets.
• Classification algorithm: Trains a support vector machine (SVM) classifier for apolipoprotein versus non-apolipoprotein discrimination.
• Performance: Achieves 97.27% accuracy, 96.30% sensitivity, and 97.76% specificity in 10-fold cross-validation for apolipoprotein identification.
• Subfamily prediction: Predicts apolipoprotein subfamilies with reported overall accuracy of 95.93%.

Scientific Applications:

• Disease mechanism investigation: Enables study of apolipoprotein involvement in hyperlipidemia, atherosclerosis, Alzheimer's disease, and diabetes.
• Target identification for drug discovery: Assists in identifying potential apolipoprotein targets for therapeutic development.

Methodology:

Computational steps include using the 270 apolipoprotein / 535 non-apolipoprotein dataset, extracting PseAAC and CKSAAP features, ranking features by ANOVA, applying incremental feature selection, training an SVM classifier, and evaluating performance via 10-fold cross-validation.