NanoStringDiff

NanoStringDiff performs differential expression analysis of NanoString nCounter count data to detect mRNA and miRNA expression changes, including data from formalin-fixed paraffin-embedded (FFPE) samples.

Key Features:

• Generalized Linear Model (GLM): Implements a GLM framework based on the negative binomial family to model overdispersed NanoString nCounter count data.
• Multifactor Experimental Design: Supports multifactor designs to incorporate and analyze multiple experimental variables simultaneously.
• Integrated Data Normalization: Uses size factors derived from positive controls and housekeeping genes and background levels from negative controls integrated into the model.
• Empirical Bayes Shrinkage: Applies empirical Bayes shrinkage to improve dispersion parameter estimation stability across genes.
• Likelihood Ratio Test: Uses a likelihood ratio test to identify differentially expressed genes from the fitted model.

Scientific Applications:

• mRNA and miRNA differential expression: Detection of differential expression for mRNA and miRNA measured by NanoString nCounter.
• FFPE and archived tissue analysis: Analysis of expression data from formalin-fixed paraffin-embedded samples for clinical and retrospective studies.
• Count-data specific analysis: Provides a statistical alternative to t-test-based methods for NanoString count data.

Methodology:

Uses a negative binomial GLM with normalization via size factors from positive controls and housekeeping genes and background from negative controls, empirical Bayes shrinkage for dispersion estimation, and likelihood ratio testing for differential expression.