AWFisher

AWFisher implements an adaptively weighted Fisher's method to combine P-values from multiple independent studies (K) and increase statistical power for omics meta-analysis.

Key Features:

• Adaptive Weighting: Employs an adaptive weighting scheme (AW-Fisher) to combine P-values from K independent studies and identify which studies contribute to signals.
• Importance Sampling with Spline Interpolation: Uses importance sampling together with spline interpolation to improve the accuracy and speed of P-value calculations for large numbers of studies.
• Variability Index via Bootstrapping: Applies bootstrapping to construct a variability index for the AW-Fisher weight estimator, quantifying the reliability of weight estimates.
• Co-membership Matrix and Gene Categorization: Constructs a co-membership matrix to categorize differentially expressed genes based on meta-patterns across studies.
• Implementation: Implemented as an R package (AWFisher) with C++ backend optimizations for computational performance.

Scientific Applications:

• Omics meta-analysis: Synthesizes P-values across numerous genomic or transcriptomic studies to detect consistent signals of differential expression.
• Differential gene expression discovery: Identifies genes that are consistently differentially expressed across independent studies.
• Biological pattern characterization: Enables clustering and interpretation of genes by meta-patterns and study-specific contributions using co-membership matrices and variability indices.

Methodology:

Combines P-values via an adaptively weighted Fisher's scheme; implements importance sampling with spline interpolation for P-value calculation; uses bootstrapping to estimate a variability index for the AW-Fisher weight estimator; constructs a co-membership matrix to categorize genes by expression meta-patterns.