STarMir

STarMir predicts microRNA (miRNA) binding sites on target RNA using logistic prediction models trained on CLIP-derived data to identify canonical (seed) and non-canonical (seedless) miRNA:target interactions.

Key Features:

• Logistic prediction models: Applies logistic prediction models trained on CLIP-derived datasets to score miRNA:target binding sites.
• CLIP-based training data: Leverages crosslinking immunoprecipitation (CLIP) data for model training and evaluation.
• Feature computation: Computes sequence, thermodynamic, and target-structure features for each predicted binding site.
• Confidence scoring: Provides logistic probability scores as quantitative measures of confidence for each predicted interaction.
• Regional and site-type coverage: Predicts both seed and seedless sites across 3' UTR, CDS, and 5' UTR regions of mRNA.
• Output content: Generates detailed outputs including binding-site features and diagrams of predicted miRNA:target hybrids.
• Cross-species validation: Models have demonstrated intra- and inter-dataset validation and general applicability across species.
• Integration with Sfold: Integrated with the Sfold RNA package for RNA structure prediction and analysis.

Scientific Applications:

• miRNA binding site identification: Identification of canonical and non-canonical miRNA binding sites on target RNAs.
• Gene regulatory network analysis: Facilitates mapping of miRNA-mediated regulation within gene regulatory networks.
• Post-transcriptional silencing studies: Supports investigation of mechanisms of post-transcriptional gene silencing by miRNAs.
• Therapeutic target discovery: Assists identification of potential miRNA-related therapeutic targets.
• RNA structure analysis: Supports RNA structure prediction and analysis through integration with the Sfold RNA package.

Methodology:

Computes sequence, thermodynamic, and target-structure features and predicts binding sites using logistic models trained on CLIP-derived datasets, reporting logistic probability scores and demonstrating intra- and inter-dataset validation for both seed and seedless interactions.