ASSA: Fast identification of statistically significant interactions between long RNAs

Area of Science:

• Computational Biology
• Genomics
• RNA Biology

Background:

• Thousands of long noncoding RNAs (lncRNAs) have been discovered in mammals.
• Some lncRNAs regulate gene expression post-transcriptionally by forming duplexes with target RNAs.
• Existing prediction tools for RNA-RNA interactions lack accuracy due to ignoring RNA secondary structure and computational limitations with long transcripts.

Purpose of the Study:

• To develop an efficient computational pipeline for predicting RNA-RNA interactions between long transcripts.
• To improve the accuracy of predicting functional RNA-RNA interactions by integrating sequence and thermodynamic approaches.
• To assess the statistical significance of predicted RNA-RNA interactions.

Main Methods:

• Developed a novel computational pipeline named ASSA.
• ASSA integrates sequence alignment with thermodynamics-based tools for analyzing long transcripts.
• Calculates hybridization strength by summing the energy of all putative duplexes.
• Implements a method for rapid estimation of statistical significance for interaction energies.

Main Results:

• ASSA efficiently predicts RNA-RNA interactions between long transcripts.
• The pipeline accurately identifies statistically significant functional hybridizations.
• ASSA demonstrated superior performance compared to 11 other tools on specific test datasets (AUC).
• Identified unique properties of [Formula: see text] repeats in human transcriptome RNA-RNA interactions.

Conclusions:

• ASSA provides an efficient and accurate method for predicting RNA-RNA interactions involving long transcripts.
• The pipeline's ability to assess statistical significance aids in identifying functional interactions.
• ASSA advances the study of lncRNA functionality and interactions within the transcriptome.