Bioinformatics insights
Why pseudobulk matters for single-cell differential expression
Cells are observations, but biological replicates remain the unit of inference. Pseudobulk analysis helps avoid pseudoreplication.
The independence problem
Cells collected from the same donor share genetics, environment, preparation, and technical effects. Treating thousands of cells as independent replicates inflates the apparent sample size and can produce overly confident p-values.
The pseudobulk strategy
Aggregate raw counts for each cell type within each biological sample, producing one profile per cell type per replicate. Standard bulk RNA-seq frameworks can then model the experimental design at the correct unit of replication.
What still needs care
Reliable results require enough biological replicates, consistent cell-type definitions, sensible minimum-cell thresholds, covariate review, and diagnostics for library size and outliers.
A defensible analysis
Report the aggregation rule, model formula, contrasts, filtering criteria, multiple-testing procedure, and sensitivity checks. Pair statistical results with effect sizes and marker-level biological validation.