staRank

staRank ranks genes from genome-wide RNA interference (RNAi) experiments to improve the stability and reproducibility of variable rankings and gene selection for downstream validation.

Key Features:

• Stability Ranking Methodology: Implements stability ranking using resampling or subsampling to produce more stable sets of genes that are more likely to be re-selected in independent validation experiments.
• Comparison and Improvement of Ranking Methods: Evaluates base ranking metrics (mean, median, t-test, and rank-sum test) and augments them with stability selection to improve reproducibility and accuracy.
• Performance Assessment: Assesses performance through simulation studies and analysis of three microscopy image RNAi datasets, finding that the rank-sum test provides an optimal balance between ranking stability and accuracy and that stability ranking enhances these attributes across base methods.
• Robustness to Limited and Noisy Data: Addresses limited sample sizes per gene knock-down and inherently noisy readouts by increasing robustness of variable rankings via resampling and stability selection.

Scientific Applications:

• Gene Prioritization from RNAi Screens: Improves selection of candidate genes from genome-wide RNAi screens for downstream validation.
• Microscopy Image-based RNAi Analysis: Enhances ranking stability and accuracy in microscopy image RNAi datasets.
• Pathway and Process Identification: Supports identification of intracellular molecular pathways and genes involved in specific biological processes.

Methodology:

Uses resampling/subsampling-based stability ranking and stability selection applied to base ranking methods (mean, median, t-test, rank-sum test), evaluated via simulation studies and application to three microscopy image RNAi datasets.