twilight

Twilight implements statistical analysis of differentially expressed genes in two-condition gene expression microarray datasets using local false discovery rate (local FDR) estimation.

Key Features:

• Local False Discovery Rate (FDR): Implements local FDR estimation to assign a probability that each gene is truly differentially expressed at specific thresholds.
• Heuristic Search Algorithm: Uses a heuristic search algorithm to estimate local FDR values efficiently across varying expression levels.
• Diagnostic Plots: Generates diagnostic plots that visualize differential expression patterns and the behavior of local FDR across the dataset.

Scientific Applications:

• Differential Expression Analysis: Identifying genes differentially expressed between two conditions (e.g., diseased vs. healthy, treated vs. untreated) in microarray datasets.
• False Discovery Control and Prioritization: Minimizing false positives and prioritizing candidate genes for downstream analysis by providing per-gene local FDR estimates.

Methodology:

Implemented as an R/Bioconductor package that estimates local FDR using a heuristic search algorithm and produces diagnostic plots, emphasizing local rather than global FDR.