iDRBP_MMC

iDRBP_MMC predicts DNA-binding proteins (DBPs) and RNA-binding proteins (RBPs) from protein sequence using sequence-based multi-label learning combined with motif-based convolutional neural networks to reduce cross-prediction and detect proteins that bind DNA, RNA, or both.

Key Features:

• Multi-Label Learning Model: Simultaneously predicts DBP and RBP labels from sequences to reduce misclassification between protein types.
• Motif-Based Convolutional Neural Network (CNN): Learns and recognizes nucleic-acid-binding sequence motifs to improve prediction accuracy.
• Sequence-Based Computational Framework: Operates on protein sequence features without requiring structural inputs.
• Cross-Prediction Reduction: Specifically designed to reduce cross-prediction between DBPs and RBPs.
• Benchmarking on Four Test Datasets: Demonstrated improved performance compared with state-of-the-art predictors across four test datasets.
• Detection of Multifunctional Binders: Identifies proteins that bind DNA, RNA, or both.

Scientific Applications:

• Large-Scale Genomic Annotation: Applied to annotate nucleic-acid-binding proteins in large genomic datasets.
• Tomato Genome Analysis: Used specifically in analysis of the tomato genome to identify DBPs, RBPs, and dual binders.
• Multifunctional Binding Protein Discovery: Enables identification and study of proteins with both DNA- and RNA-binding activities.

Methodology:

Multi-label learning integrated with a motif-based convolutional neural network applied to sequence-based inputs.