JDINAC

JDINAC detects differential gene interaction networks using joint density-based non-parametric methods to identify network biomarkers and classify sample groups while capturing non-linear relationships in high-dimensional sparse omics data.

Key Features:

• Non-Parametric Approach: Employs non-parametric methods to capture non-linear relationships between molecular interactions without assuming parametric distributions, suitable for high-dimensional sparse omics data.
• Density-Based Analysis: Utilizes joint density-based techniques to identify differential interaction patterns within networks across different groups (e.g., diseased vs. healthy).
• Adjustment for Confounding Factors: Incorporates adjustment for potential confounding variables to improve reliability and accuracy of network estimation.
• Network Biomarkers for Classification: Leverages identified differential interactions as network biomarkers to construct classification models that distinguish sample groups.

Scientific Applications:

• Disease Mechanism Elucidation: Compares networks from diseased and healthy samples to aid in uncovering underlying mechanisms of pathogenesis.
• Biomarker Identification: Identifies network biomarkers that can serve as potential clinical indicators for disease status.
• Feature Selection and Classification: Provides a framework for feature selection from high-dimensional sparse omics data to facilitate accurate classification models.
• Breast Invasive Carcinoma Analysis: Applied to a Breast Invasive Carcinoma dataset comprising 114 patients with tumor and matched normal samples to identify hub genes and interaction patterns and to discriminate tumor versus normal samples.

Methodology:

Analyzes gene interaction networks between two groups using joint density-based non-parametric methods that capture non-linear relationships and adjust for confounding variables, and uses identified differential interactions as biomarkers for classification.