PIANO

PIANO predicts transcriptome-wide pseudouridine (Ψ) sites and annotates them with post-transcriptional regulatory information to support analysis of pseudouridylation distribution and potential regulatory mechanisms.

Key Features:

• Prediction framework: Integrates conventional sequence features with 42 additional genomic attributes to predict pseudouridylation sites genome-wide.
• Model types: Provides separate full transcript models and mature mRNA models for Ψ-site prediction.
• Performance: Achieved average AUCs of 0.955 for full transcript models and 0.838 for mature mRNA models across six independent datasets.
• Benchmark datasets and technologies: Evaluated against six independent datasets derived from four Ψ-profiling technologies: Ψ-seq, RBS-seq, Pseudo-seq, and CeU-seq.
• Comparative improvement: Demonstrated significant improvement over prior in silico methods PPUS, iRNA-PseU, and PseUI.
• Regulatory annotation: Systematically annotates predicted Ψ sites with associations to miRNA targets, RNA-binding protein (RBP) binding regions, and splicing sites.
• Experimental site database: Includes a collection of 4,303 experimentally validated human Ψ sites, representing the largest compiled dataset reported.

Scientific Applications:

• Transcriptome-wide mapping: Identification and mapping of pseudouridylation sites across transcriptomes.
• Functional hypothesis generation: Linking predicted Ψ sites to miRNA, RBP, and splicing annotations to generate hypotheses about regulatory mechanisms.
• Prioritization for validation: Prioritization of candidate Ψ sites for experimental validation using prediction outputs and associated annotations.
• Cross-technology comparison: Comparison of predictions with data from Ψ-seq, RBS-seq, Pseudo-seq, and CeU-seq to assess concordance with experimental profiling.

Methodology:

Integrates conventional sequence features with 42 genomic attributes to build full transcript and mature mRNA predictive models and evaluates them using six independent datasets derived from Ψ-seq, RBS-seq, Pseudo-seq, and CeU-seq.