miRcorrNet

miRcorrNet integrates microRNA (miRNA) and messenger RNA (mRNA) expression profiles using machine learning to identify correlated miRNA–mRNA groups and prioritize regulators relevant to disease mechanisms such as cancer.

Key Features:

• Machine Learning-Based Integration: Employs a machine learning framework to integrate microRNA (miRNA) and messenger RNA (mRNA) expression profiles and infer functional miRNA–mRNA relationships.
• Feature Grouping and Ranking: Groups mRNAs by correlation with miRNA expression into gene clusters and ranks these groups to classify and prioritize candidates.
• Comparative Evaluation on TCGA: Evaluated on The Cancer Genome Atlas (TCGA) with reported area under the curve (AUC) exceeding 95% in comparative analyses.
• Independent Dataset Validation: Validated using independent datasets to assess robustness across different experimental conditions.
• Biological Function Exploration: Performs literature-search-based validation of identified miRNAs against known databases, reporting approximately 90% validation accuracy.
• Ranking for Case-Control Separation: Incorporates ranking steps aimed at facilitating separation of case and control samples for differential analysis.

Scientific Applications:

• PAN-CANCER miRNA Prioritization: Prioritizes high-confidence miRNAs that regulate pan-cancer pathways for oncology research.
• Integrative Transcriptomic Analysis: Supports discovery of molecular mechanisms by integrating miRNA and mRNA expression to identify regulatory relationships in disease studies.

Methodology:

Operates within the KNIME analytics platform using ".table" extension files; requires matched mRNA and miRNA expression profiles with control-case columns; employs machine learning-based integration, groups mRNAs by correlation with miRNA expression, ranks gene groups to classify and separate case and control samples, and evaluates results on TCGA and independent datasets with literature-search-based validation.