scanMiR

scanMiR implements a biochemically based toolkit for versatile and efficient microRNA target prediction, translating a biochemical model combined with empirically-derived k-mer affinity predictions into a flexible Bioconductor package that efficiently scans for binding sites, estimates their affinity, and predicts aggregated transcript repression.

Key Features:

• miRNA affinity model support: Works with KdModels for miRNA binding affinity predictions.
• Multiple scanning modes: Supports scanning using miRNA seeds, full miRNA sequences enabling 3' alignment, and KdModels.
• 3'-supplementary alignment: Flexible 3' alignment enables prediction of unconventional interactions.
• Slicing site prediction: Predicts binding sites potentially leading to slicing.
• TDMD site prediction: Predicts target-directed microRNA degradation sites.
• Transcript repression prediction: Predicts aggregated transcript repression using lightweight linear models.
• Visualization functions: Includes utility and plotting functions for visual representation of miRNA-target alignment.

Scientific Applications:

• Conventional miRNA target prediction: Identifies functionally relevant miRNA targets through biochemically-based affinity modeling.
• Unconventional interaction detection: Predicts bindings potentially leading to target-directed microRNA degradation or slicing.
• Neuronal transcript analysis: Enables systematic scanning for unconventional sites on neuronal transcripts including lncRNAs and circRNAs.

Methodology:

scanMiR uses lightweight linear models combined with a biochemical model and empirically-derived k-mer affinity predictions to scan for binding sites, estimate affinity, and predict aggregated transcript repression. The approach enables flexible 3'-supplementary alignment for predicting unconventional interactions.