HAllA

HAllA applies a hierarchical all-against-all association testing framework to detect linear and non-linear block-wise relationships between paired high-dimensional multi-omic datasets while controlling the false discovery rate.

Key Features:

• Hierarchical Framework: Utilizes a hierarchical approach to organize features and perform structured association testing between paired datasets.
• False Discovery Rate Correction: Integrates false discovery rate correction within hierarchical hypothesis testing to provide statistical control over multiple comparisons.
• Block-wise Relationship Detection: Detects significant linear and non-linear block-wise relationships among continuous and/or categorical data types.
• Optimization and Evaluation: Optimized and evaluated on heterogeneous synthetic datasets with known association structures, showing superior performance compared to traditional all-against-all and other block testing approaches across various similarity measures.
• Feature Similarity-based Grouping: Leverages feature similarity within each dataset to identify statistically significant groups of features across datasets, improving sensitivity and interpretability.

Scientific Applications:

• Gene Expression and Host Immune Activity: Identifies significant relationships between gene expression profiles and host immune activity.
• Microbiome and Host Transcriptome: Discovers interactions between microbiome composition and host transcriptomic data.
• Metabolomic Profiling: Links metabolomic profiles with various biological processes and conditions.
• Human Health Phenotypes: Explores associations between multi-omic data and human health-related phenotypes to inform potential biomarkers or therapeutic targets.

Methodology:

Performs hierarchical all-against-all association testing using feature similarity to group features, applies hierarchical hypothesis testing with false discovery rate correction, and detects linear and non-linear block-wise relationships.