TargetDBP+

TargetDBP+ identifies DNA-binding proteins to improve detection and characterization of protein–DNA interactions for genomic and regulatory studies.

Key Features:

• Feature Extraction: Extracts five feature matrices: Amino Acid One-Hot Matrix (AAOHM), Position-Specific Scoring Matrix (PSSM), Predicted Secondary Structure Probability Matrix (PSSPM), Predicted Solvent Accessibility Probability Matrix (PSAPM), and Predicted Probabilities of DNA-Binding Sites (PPDBSs).
• Weighted Convolutional Features: Combines extracted features in a weighted, serial manner and learns element-wise weights using the differential evolution algorithm.
• Model Training: Trains a support vector machine (SVM) classifier on the optimized features for DBP prediction.
• Benchmark Dataset: Uses the UniSwiss gold-standard dataset derived from UniprotKB/Swiss-Prot comprising 4,881 DBPs and 4,881 non-DBPs for evaluation.
• Performance Metrics: Demonstrated accuracy of 85.83%, precision of 88.45%, independent-validation coverage of 82.41%, and a Matthews Correlation Coefficient (MCC) of 0.718.

Scientific Applications:

• DBP Identification: Improves identification of DNA-binding proteins from protein sequence data.
• Gene Regulation Analysis: Supports analysis of protein–DNA interactions relevant to gene regulation mechanisms.
• Disease Pathway Investigation: Aids investigation of disease-related protein–DNA interactions and potential therapeutic targets.
• Genomic Research: Applicable to genomic studies requiring systematic characterization of protein–DNA binding.

Methodology:

Extracts AAOHM, PSSM, PSSPM, PSAPM, and PPDBSs features, combines them in a weighted serial manner with weights learned by the differential evolution algorithm, trains a support vector machine (SVM) classifier on the optimized features, and evaluates performance on the UniSwiss dataset derived from UniprotKB/Swiss‑Prot (4,881 DBPs and 4,881 non-DBPs).