Pprint

Pprint predicts RNA-binding residues in protein sequences to identify RNA–protein interaction sites relevant to post-transcriptional gene regulation.

Key Features:

• Prediction target: Predicts RNA-binding residues within protein sequences (RNA-binding proteins, RBPs).
• Input features: Uses evolutionary information in the form of Position-Specific Scoring Matrix (PSSM) profiles derived from multiple sequence alignments as model input.
• Algorithm: Implements support vector machine (SVM) classifiers to predict residue-level RNA binding.
• Evaluation metric: Reports predictive performance using Matthews correlation coefficient (MCC), with MCC=0.45 for PSSM+SVM on the primary dataset and MCC=0.32 on an alternative dataset.
• Benchmark comparison: Achieves higher MCC (0.45) than a prior maximum MCC of 0.41 on the same dataset.
• Validation datasets: Trained and tested on datasets comprising 86 RNA-binding protein chains and alternatively on 107 RBP chains.

Scientific Applications:

• Post-transcriptional gene regulation: Identifies residue-level RNA contacts to inform studies of mRNA processing and regulatory complexes.
• Protein–RNA interaction mapping: Guides experimental and computational mapping of protein–RNA interfaces at the residue level.
• Developmental biology: Supports investigation of RBPs involved in gene expression control during organismal development.
• Genomics and molecular medicine: Aids interpretation of variants and functional annotation of proteins implicated in disease via altered RNA binding.

Methodology:

Generates PSSM profiles from multiple sequence alignments as input to SVM classifiers; models were trained and tested using fivefold cross-validation on a dataset of 86 RBP chains and validated on an alternative dataset of 107 RBP chains, with performance measured by Matthews correlation coefficient (MCC).