RactIP

RactIP predicts RNA-RNA interactions by modeling complex joint secondary structures, including kissing hairpins, using integer programming to integrate probabilistic base-pairing information.

Key Features:

• MEA-based prediction: Employs a maximum expected accuracy (MEA) structure approach formulated as an integer programming (IP) problem with threshold cuts.
• Probabilistic integration: Incorporates posterior internal and external base-pairing probabilities derived from an ensemble of equilibrium joint structures into its objective function.
• Kissing hairpin modeling: Supports prediction of complex joint secondary structures including kissing hairpins.
• General interaction types: Applicable to a broad range of RNA-RNA interaction classes rather than a single specific class.
• Computational efficiency: Formulated to run faster than many existing methods while aiming to maintain comparable prediction accuracy.

Scientific Applications:

• Gene Regulation: Identifies potential targets of non-coding small RNAs to investigate post-transcriptional gene regulation mechanisms.
• Structural Biology: Infers complex joint secondary structures to support functional and structural analyses of RNA molecules.
• Comparative Studies: Enables comparison of predicted interaction sites with experimental data for validation and further exploration of RNA functions.

Methodology:

Implemented in C++, RactIP integrates posterior internal and external base-pairing probabilities from an ensemble of equilibrium joint structures into an integer programming framework, applies a maximum expected accuracy (MEA) objective with threshold cuts, and predicts general RNA-RNA joint secondary structures including kissing hairpins.