We further partitioned iPSC gene expression variation using the Tier 1 expression array data, the assay with the largest number of donors and lines. Of the 25,434 probes analysed (16,829 genes) (Supplementary Table 3), donor effects explained the largest proportion of variation in 46.4% of probes (53.3% of genes), substantially more than any other factor, including copy number status (23.4%), culture conditions (26.2%), passage (2%) and gender (1.9%, Fig. 3d). Donor effects were common, and consistent across large numbers of genes, while others such as CNA status had larger effects on a smaller number of genes (Fig. 3d). We observed minor effects of gender and line passage number on RNA-seq, methylation and protein immunofluorescence (Fig. 3d, Extended Data Fig. 6). Likewise, we did not observe substantial changes in PluriTest scores, or pluripotency marker expression across passages (P > 0.3, Extended Data Fig. 6), reflecting that pluripotency was maintained during culture. In principle, the estimated donor variation could arise due to shared reprogramming environment because lines were derived from the same population of fibroblast cells. However, expression quantitative trait locus (eQTL)
