IDRBP-PPCT

IDRBP-PPCT predicts nucleic acid-binding proteins (NABPs), including DNA-binding proteins (DBPs), RNA-binding proteins (RBPs), and dual RNA–DNA binding proteins (DRBPs), from protein sequences to support analysis of protein–nucleic acid interactions involved in replication, transcription, and translation.

Key Features:

• PPCT representation: Position-Specific Scoring Matrix (PSSM) and Position-Specific Frequency Matrix (PSFM) Cross Transformation (PPCT) captures evolutionary information embedded within PSSMs and PSFMs and their correlations.
• Fixed-dimension feature vectors: PPCT transforms protein sequences into fixed-dimension feature vectors suitable for computational analysis.
• Two-layer random forest framework: A two-layer predictive framework based on the random forest algorithm integrates PPCT-derived features for classification of DBPs, RBPs, and DRBPs.
• Validation: Model performance was validated using independent datasets and the tomato genome.

Scientific Applications:

• Protein class identification: Identification of DBPs, RBPs, and DRBPs from protein sequences.
• Protein–nucleic acid interaction analysis: Investigation of interactions between proteins and nucleic acids relevant to gene expression processes.
• Gene expression studies: Analysis of proteins involved in replication, transcription, and translation.
• Genome-wide annotation: Genome-scale prediction and annotation of nucleic acid-binding proteins, as applied to the tomato genome.

Methodology:

PPCT is applied to PSSMs and PSFMs to produce fixed-dimension feature vectors, which are input to a two-layer random forest classifier; performance was evaluated on independent datasets and the tomato genome.