We transduced primary human astrocytes (fetal cortex-derived) from three donors with an all-in-one CRISPR activation lentivirus (VP64-dSpCas9-VP64) carrying either gRNA-hit, a guide designed against the FOXO4 promoter, or a non-targeting guide. The cells were then cultured under a neuronal reprogramming protocol and harvested on day 10 after transduction; all samples come from that one time point, so there is no time variable.

The guide is in the `group` column: gRNA-hit means CRISPR activation with the FOXO4-promoter guide, and non-targeting means the non-targeting control guide; the `geo_genotype` column repeats the same labels. Each sample is one astrocyte culture from one donor under one guide. The donor is in the `donor` column (donor1, donor2, donor3, matching the number at the end of each sample name): each donor contributed one gRNA-hit culture and one non-targeting culture, so the two samples of a donor come from the same donor's cells. Each library was sequenced once; there are no technical replicates.

We have no record of processing or sequencing batches and no other variable we consider relevant; the remaining columns are GEO annotations. The data are bulk RNA-seq, given as gene-level raw read counts from featureCounts (Gencode v22 annotation).

Our question is which genes are differentially expressed in gRNA-hit astrocytes compared with non-targeting control astrocytes on day 10.