Fig. 1

Data quality control metrics. a Representative images of hiPSC-NPCs (left) and 6-week-old forebrain neurons (right) from control (top) and PMS probands (bottom). hiPSC-NPCs stained with PAX6 (red), NESTIN (green); hiPSC-neurons stained with MAP2 (green), DAPI-stained nuclei (blue). Pairwise correlations compared (b) hiPSC-NPC and (c) hiPSC-neuron transcriptomes from the same clone and same induction (n = 12, n = 31, respectively), same clone but different induction (n = 46, n = 55, respectively), all related family members (n = 68, n = 57, respectively) and all unrelated family members (n = 505, n = 677, respectively). Analysis of variance for multiple comparisons was used to test for differences between the means of correlation coefficients. d Linear mixed modelling was used to compute the percentage of gene expression variance explained according to six factors, which represent potential biological sources of variability. Differences in cell types and donor as a repeated measure, followed by excitatory neuron cell composition (estimated using CiberSort in grey) explains the largest amount of variability in the transcriptome data. e Principal components analysis of gene expression data from hiPSC-NPCs (red) and hiPSC-neurons (blue), each unique shape denotes one specific donor. Note, there was no distinct stratification by PMS case status based on global expression profiles. (f) Genes that vary most across donors are enriched for brain cis-eQTLs. Fold enrichment (log₂) for the 2000 top cis-eQTLs discovered in post mortem dorsolateral prefrontal cortex data generated by the CommonMind Consortium shown for six sources of variation, plus residuals. Each line indicates the fold enrichment for genes with the fraction of variance explained exceeding the cutoff indicated on the x-axis. Enrichments are shown on the x-axis until less than 100 genes pass the cutoff