We studied mouse splenic NK cells after CRISPR editing with either a Fli1-targeting guide (Fli1 knockout) or a control guide (NTC). The `group` column, repeated in `treatment` and `geo_treatment`, holds the condition: Fli1 KO (three libraries) or NTC (two libraries). All cells were harvested at a single time point after editing, so there is no time variable.

Each library is one edited bulk NK-cell population, and there are no technical replicates. Donor mice are not recorded, and whether a library came from one spleen or from several pooled spleens is not stated. The sample names number the libraries within each condition (Fli1 KO Rep1 to Rep3, NTC Rep1 and Rep2), but there is no record that same-numbered knockout and control libraries share a donor or an editing run, and the third knockout library has no control with its number. We have no record of processing or sequencing batches, and no other variable matters to us; the remaining columns are GEO annotations.

The data are bulk total RNA-seq (Illumina RNA Prep with Enrichment). The matrix holds raw gene-level read counts after STAR alignment to mm10, one row per Ensembl gene ID.

Our question is which genes are differentially expressed in Fli1 knockout NK cells (Fli1 KO) compared with control-guide NK cells (NTC), using all five libraries.