MechRNA

MechRNA predicts mechanisms of long non-coding RNAs (lncRNAs) by integrating RNA-RNA and RNA-protein interaction predictions to infer mechanistic hypotheses of lncRNA action.

Key Features:

• RNA–RNA interaction prediction: Uses an enhanced version of IntaRNA2 optimized for efficient prediction with very long input sequences and large-scale analyses.
• RNA–protein interaction integration: Incorporates pre-computed protein binding data generated with GraphProt for both the lncRNA and its target transcripts.
• Mechanism generation: Produces candidate mechanisms based on predicted interactions, relative interaction locations, and correlation data.
• Mechanism scoring: Selects the most likely mechanistic explanation using a combined P-value approach.
• Transcriptome-wide target prediction: Enables transcriptome-wide analysis and identification of high-confidence targets per lncRNA.

Scientific Applications:

• Cancer lncRNA mechanism discovery: Applied to cancer-related lncRNAs including PCAT1, PCAT29, ARLnc1, PCA3, and 7SL to predict mechanisms and targets.
• Competitive binding identification: Identified competitive binding mechanisms such as the interaction between 7SL and HuR on the tumor suppressor TP53.
• Target prioritization: Identified hundreds of high-confidence potential targets for individual lncRNAs.
• Experimental validation support: Predictions guided validation of specific interactions, exemplified by the PCAT1-BRCA2 interaction confirmed with HuR immunoprecipitation assays.

Methodology:

Computational steps include RNA–RNA prediction with an enhanced IntaRNA2 for long sequences, incorporation of pre-computed GraphProt protein-binding data for lncRNAs and targets, generation of candidate mechanisms from predicted interactions, locations and correlation data, and selection of mechanisms via a combined P-value approach.