stem cell and neural progenitor markers (e.g., VIM, SMAD2 and NOTCH2). These suggest that variation in the degrees of neuron differentiation and maturation existed in our samples and needs to be accounted for in the differential expression analysis. Note that all of our samples showed a characteristic expression pattern for neuronal samples (higher expression of neuronal markers and lower expression of NPC markers) when compared with the NPC samples derived from a subset of our control iPSCs previously [42] (data not shown). Therefore, we considered the two clusters as two “batches” and applied ComBat [43] (a batch-correction tool adjusting for differences in the means across the batches and the variances, which would not be considered in a standard two-factor analysis) to correct the raw gene expression values and used the “batch”-corrected values for all subsequent analysis. Note that heterogeneity in neural induction is a rather common issue that has been discussed previously not only for our protocol [26], but also for others [44, 45], and the usage of multiple iPSC clones/lines from the same individuals has been suggested [46–48].
