PlDBPred

PlDBPred predicts DNA-binding proteins (DBPs) in plants to provide plant-specific identification of DBPs across species and address limitations of models trained primarily on human and mouse datasets.

Key Features:

• Plant DBP prediction: Predicts DNA-binding proteins (DBPs) across plant species.
• Learning approaches: Implements and compares shallow learning and deep learning approaches, including five shallow learning algorithms and six deep learning models.
• Best-performing algorithm: Support vector machine (SVM) was identified as the best-performing model.
• Cross-validation performance: SVM achieved 94.0% AUC-ROC and 93.5% AUC-PR in repeated 5-fold cross-validation.
• Independent validation: Achieved 93.8% AUC-ROC and 94.6% AUC-PR on an independent dataset.
• Comparative performance: Outperformed existing state-of-the-art tools in accuracy for plant DBP prediction.
• Biological focus: Targets proteins involved in nucleotide recognition, transcriptional control, and gene expression regulation.

Scientific Applications:

• Plant DBP identification: Enables identification of DNA-binding proteins in plant genomes and proteomes.
• Functional genomics: Supports analysis of transcriptional regulators and proteins involved in nucleotide recognition and gene expression regulation.
• Cross-species prediction: Improves DBP prediction applicability beyond Arabidopsis to other plant species.

Methodology:

Compared five shallow learning algorithms and six deep learning models, selected support vector machine (SVM) as the best model, evaluated with repeated 5-fold cross-validation and validated on an independent dataset.