SFDR

SFDR computes stratified False Discovery Rate (FDR) q-values for genome-wide Single Nucleotide Polymorphism (SNP) association analyses, enabling control or estimation of FDR within strata defined by genome-wide linkage scan results and via weighted schemes.

Key Features:

• Stratified FDR Control: Computes and controls or estimates FDR separately within predefined strata to leverage natural stratifications in genetic hypothesis testing.
• Utilization of Genome-Wide Linkage Scan Results: Accepts strata assignments derived from genome-wide linkage scan results for stratified analysis.
• Weighted FDR q-values: Computes weighted FDR q-values as an alternative to unweighted stratified FDR estimation.
• Fixed Rejection Region Framework: Implements a framework that rejects all hypotheses with unadjusted p-values below a prespecified threshold and estimates overall FDR as a weighted average of stratum-specific FDRs.
• Fixed FDR Framework: Implements a framework that rejects as many hypotheses as possible while controlling the overall FDR at a predetermined level and specifies a condition related to expected total true positives versus aggregated methods.

Scientific Applications:

• Genome-wide association (GWA) studies: Applies stratified FDR control to genome-wide SNP association analyses to manage multiplicity and false positives in high-throughput genotyping data.
• Power enhancement in stratified analyses: Enables identification of more true positives within strata while maintaining control over FDR, particularly when conventional low-rate FDR control (e.g., 5% or 10%) may be infeasible.

Methodology:

Computes stratum-specific FDR q-values and supports two explicit frameworks: the Fixed Rejection Region framework, which rejects hypotheses with unadjusted p-values below a preset threshold and estimates overall FDR as a weighted average of stratum-specific FDRs; and the Fixed FDR framework, which rejects as many hypotheses as possible while controlling the overall FDR at a predetermined level and specifies a condition regarding expected total true positives relative to aggregated methods.