We transitioned naive human pluripotent stem cells of the line HNES1 toward a primed-like state under one of three treatments: the EED PROTAC degrader UNC6852 (2.5 µM), the EZH1/2 catalytic inhibitor UNC1999 (1 µM), or DMSO vehicle. The drugs were added 24 hours before the transition started (day 0) and renewed daily, and all cultures were harvested on day 5 of the transition, so all samples share one time point and there is no time variable. The treatment is in the `group` column: PROTAC for UNC6852, UNC for UNC1999 and DMSO for the vehicle; the `treatment` column repeats the same labels.

Each sample is one HNES1 culture under one treatment. All nine come from the same cell line, so there are no donors, and no two samples come from the same culture. The whole experiment was performed independently three times, and each experiment included one DMSO, one PROTAC and one UNC culture. The `replicate` column (Rep1, Rep2, Rep3, taken from the end of the sample names) records which experiment a culture belongs to, so the three experiments are three experimental batches. Each library was sequenced once; there are no technical replicates.

Apart from the experiments, 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 raw directional read counts per gene from SeqMonk (Ensembl v113 annotation).

Our question is which genes are differentially expressed in PROTAC-treated cultures compared with DMSO-treated cultures on day 5. We would like to use all nine samples, including the UNC1999 cultures from the same experiments.