iDBP-DEP

iDBP-DEP predicts DNA-binding proteins from protein sequences by extracting discriminative features and applying feature selection to improve sequence-based identification.

Key Features:

• Multi-view Feature Integration: Integrates discriminative features derived from evolutionary profiles, dipeptide compositions, and physicochemical properties of proteins.
• Feature Selection Mechanism: Applies an advanced feature selection process to retain the most relevant features for the prediction model.
• Robust Evaluation Methodology: Evaluates performance using the Jackknife test on benchmark datasets PDB1075 and PDB594, reporting improvements of 1.8% and 3.0% in accuracy (Acc) and Matthew's Correlation Coefficient (MCC) respectively on PDB1075, and 7.4% and 14.8% improvements in Acc and MCC on PDB594.
• Independent Validation: Validates predictions on an independent dataset PDB186, achieving an accuracy of 80.1% and an MCC of 0.684.

Scientific Applications:

• Gene regulation studies: Identification of DNA-binding proteins to support analyses of gene regulatory mechanisms.
• Genetic network and disease mechanism research: Detection of DNA-binding proteins implicated in regulatory networks relevant to disease studies.
• Therapeutic target identification: Support for developing therapeutic strategies targeting specific protein–DNA interactions.

Methodology:

Extracts discriminative features from evolutionary profiles, dipeptide compositions, and physicochemical properties, applies feature selection to refine inputs, and evaluates predictions via Jackknife testing on PDB1075 and PDB594 with independent validation on PDB186.