PredPRBA

PredPRBA predicts protein–RNA binding affinity using Gradient Boosted Regression Trees (GBRT) trained on 37 sequence and structural features.

Key Features:

• Gradient Boosted Regression Trees (GBRT): Employs GBRT to model and predict binding affinity, leveraging its ability to handle complex datasets and capture nonlinear relationships.
• Comprehensive Feature Set: Generates 37 sequence and structural features that capture multiple aspects of protein–RNA interactions, with emphasis on RNA structural properties.
• Categorical Analysis: Categorizes protein–RNA complexes by RNA type to enable tailored modeling for different interaction classes.
• Cross-Validation Evaluation: Evaluates predictive performance using leave-one-out cross-validation and benchmarks results against traditional regression methods and SPOT-Seq-RNA.
• High Correlation Coefficients: Reports correlation coefficients ranging from 0.723 to 0.897 across different RNA categories.

Scientific Applications:

• Molecular Biology: Supports studies of protein–RNA recognition mechanisms and identification of strong binding partners.
• Structural Bioinformatics: Provides quantitative affinity predictions for analysis and interpretation of protein–RNA complex structures.
• Drug Discovery: Supplies binding affinity estimates to assist prioritization of protein–RNA interactions for therapeutic targeting.

Methodology:

Constructed a dataset from 103 manually curated protein–RNA complex structures from the literature, generated 37 sequence and structural features, built GBRT models for each RNA category, and evaluated performance with leave-one-out cross-validation and comparison to traditional regression methods and SPOT-Seq-RNA.