Chapter 5 Differential expression, FDR, GO, and GSEA

DESeq2 is a popular and accurate computational algorithm to detect differential gene expression from RNA-seq data. It includes many elegant quantitative considerations, such as:

  • Normalize the gene read counts by library size and composition
  • Model gene read counts with negative binomial distribution
  • Use hierarchical modeling to stabilize the gene variance
  • Use Benjamini-Hochberg to calculate control for false discovery rate of calling differentially expressed genes
  • Filter lowly expressed genes to reduce the number of hypotheses to be tested

5.1 DESeq2 library normalization

5.2 DESeq2 variance stabilization

5.3 Multiple hypotheses testing and False Discovery Rate

5.4 DESeq2 gene filtering

5.5 Gene Ontology (GO analysis)

5.6 Gene Set Enrichent Analysis (GSEA)

5.7 DESeq2 tutorial