We asked whether the 49 genes differentially expressed between our isogenic hESCs and hiPSCs are also dysregulated in hiPSC lines derived from primary somatic cells as well as in other published datasets. First, we reanalyzed a published set of unmatched hESC (n=18) and hiPSC (n=12) lines generated from primary fibroblasts using retroviral vectors, whose gene expression patterns were previously analyzed by microarrays6. Since many of the 49 DEGs were not covered in the available microarray data, we performed RNA-sequencing of these hESC and hiPSC lines, which offers increased sensitivity, especially for low-abundance transcripts35,36. However, unsupervised clustering was unable to separate these hESCs from hiPSCs (Fig. 5A). Although 3 DEGs (RP11-1, MEG3, AL1327) were identified between unmatched hESCs and hiPSCs, these were likely false positives based on permutation analysis. Indeed, supervised clustering of all samples with these 3 DEGs (data not shown) or an extended set of 16 DEGs using loosened criteria could not distinguish hESCs from hiPSCs (Fig. 5A and Supplementary Fig. 4A,D). Notably, our stringently defined 49-DEG signature was also unable to segregate the transcriptomes of this extended set of hESC and hiPSC lines (Fig. 5B).
