Differential Expression

• Once expression is normalised between samples, testing for between-group differences is in principle straightforward, and t-tests have been used for this purpose, but two factors complicate the analysis:
  • Over-dispersion of variance
    • usually accounted for using a negative binomial distribution
  • Confounding factors
    • usually accounted for with a linear modeling framework
• DESeq2: Likelihood ratio tests or Wald tests
• EdgeR: Liklihood ratio tests