Interaction Model: ~genotype * treatment
Advanced ModelsTest whether the effect of one factor on gene expression depends on the level of another factor by including an interaction term in the design formula.
When to Use
- You suspect the treatment effect differs between genotypes, cell lines, sexes, or other grouping variables.
- Your biological hypothesis is specifically about the interaction -- for example, "the drug works in WT but not in KO mice".
- You have a full factorial design with sufficient replicates in every cell of the design matrix.
Required Inputs
- At least two categorical factors available in the metadata.
- Interaction runs configured through the factor selectors (easyCris builds the ~A * B formula automatically).
- Reference levels set for both factors so the contrast direction is unambiguous.
What to Expect
- easyCris generates main-effect contrasts and interaction contrasts for the selected factors.
- The interaction term (e.g., genotype:treatment) captures genes where the treatment fold-change differs across genotype levels.
- A significant interaction term means the magnitude or direction of the treatment effect is not the same in all genotype groups.
Common Pitfalls
- Interaction models require more replicates than main-effect-only models. Aim for at least three biological replicates per cell of the factorial design.
- A non-significant interaction does not mean the main effects are absent -- it means the main effects are consistent across factor levels.
- When either factor has three or more levels, the interaction generates multiple contrasts. Check all of them rather than relying on a single summary.
- Confounded designs where one factor level is nested entirely within another make the interaction effect unestimable.
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.