Roleswitch

Roleswitch infers probabilistic miRNA-mRNA interaction signatures (ProMISe) from paired miRNA and mRNA expression data to characterize miRNA-mediated gene regulation.

Key Features:

• Probabilistic Inference: Uses a probabilistic model (ProMISe) that accounts for the competitive nature of miRNA-mRNA binding to infer interactions.
• Expression Data Utilization: Operates on paired expression data from single samples of miRNAs and mRNAs.
• Iterative Updating Mechanism: Iteratively updates inferred total transcription levels of mRNA targets based on their probability of being targeted by miRNAs.
• Seed Matching Consideration: Incorporates a probabilistic function that evaluates expressed seed matches (miRNA binding sites) between miRNAs and mRNAs.

Scientific Applications:

• Cancer Diagnostics: Has demonstrated superior diagnostic power compared to using expression profiles alone for identifying biomarkers and therapeutic targets.
• Gene Set Enrichment Analysis: Facilitates identification of target enrichments specific to oncomirs such as miR-21 and miR-145 in glioblastoma and ovarian cancers, respectively.
• Tumor-Specific Interactions: Compares matched tumor/normal samples to uncover tumor-related interactions, including networks involving miR-155/183/21 in breast cancer (BRCA) and miR-221/222 in thyroid cancer (THCA).

Methodology:

Applies a probabilistic framework that integrates paired miRNA and mRNA expression data with sequence-derived expressed seed matches, models competition among multiple mRNAs for the same miRNAs, assumes observed mRNA levels are reduced relative to total transcription due to miRNA-mediated regulation, and iteratively refines inferred total transcription levels and target probabilities.