RIsearch

RIsearch predicts RNA-RNA interactions to enable rapid genome-wide identification of regulatory RNA duplexes such as bacterial sRNA–mRNA and eukaryotic miRNA–mRNA interactions.

Key Features:

• Simplified Turner energy model: Employs a simplified version of the Turner energy model to approximate RNA hybridization energies for fast computation of near-complementary duplexes.
• Smith-Waterman-like algorithm: Uses an algorithm similar to Smith-Waterman with a dinucleotide scoring matrix that mirrors Turner nearest-neighbor energies.
• Performance and speed: Achieves at least a 2.4× increase in speed for searching near-complementary regions compared to RNAplex.
• Accuracy: Maintains prediction accuracy comparable to RNAplex, validated on bacterial sRNA–mRNA and eukaryotic miRNA–mRNA benchmark datasets.
• Pre-filtering capability: Acts as a pre-filter in genome-wide screens, reducing candidate binding sites reported by TargetScanS and miRanda by up to 70% and effectively filtering bacterial RNA-RNA interaction data.

Scientific Applications:

• Genome-wide interaction screening: Rapidly identifies candidate RNA-RNA duplexes across genomes for large-scale interaction mapping.
• Non-coding RNA target discovery: Supports prediction of sRNA and miRNA targets to study regulatory roles of non-coding RNAs.
• Pre-filtering for target prediction pipelines: Reduces candidate sets from miRNA target prediction programs (TargetScanS, miRanda) to focus downstream analyses.

Methodology:

Computational approach based on a simplified Turner energy model combined with a Smith-Waterman-like algorithm using a dinucleotide scoring matrix that mirrors Turner nearest-neighbor energies for searching near-complementary regions.