Skip to content

Multi-factor Model: ~condition + batch

Model Configuration

Add additional factors or continuous covariates to the design formula so that confounding variables are controlled while testing your primary factor of interest.

When to Use

  • Your samples were processed in different batches, on different sequencing lanes, or at different sites and you need to account for this technical variation.
  • You want to adjust for a continuous covariate such as RNA integrity number (RIN), patient age, or library size.
  • PCA of your count data shows separation driven by batch rather than by your experimental factor.

Required Inputs

  • The primary factor of interest plus one or more additional effects or covariates selected from metadata columns.
  • The design preview (e.g., ~batch + condition) updates dynamically from your selections.
  • Categorical covariates must have two or more levels; continuous covariates must contain numeric values.

What to Expect

  • easyCris adjusts for all configured covariates before estimating the fold change for your primary contrast.
  • Results reflect the condition effect after accounting for batch or other confounders, which often increases the number of significant genes.
  • Adding a meaningful batch term typically increases statistical power by absorbing unexplained variance.

Common Pitfalls

  • Confounded designs -- where all treated samples are in Batch 1 and all controls are in Batch 2 -- cannot separate the treatment effect from the batch effect. The model cannot resolve this.
  • Including too many covariates relative to sample size causes overfitting. Keep the model parsimonious: include only covariates with biological or technical justification.
  • Covariates with very low variation across samples add noise without improving the model and can make fitting unstable.

Citations

References

  • Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15, 550.
  • Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B, 57(1), 289-300.