Fig. 4: DLG2 regulates a cascade of transcriptional programs driving neurogenesis & differentiation.

a Enrichment for common schizophrenia risk variants in genes up- & down-regulated between each successive pair of timepoints in wild-type (WT), conditioning on all WT-expressed genes. Dotted line indicates P_corrected = 0.05 following Bonferroni correction for 6 tests. b Four discrete transcriptional programs initiated following the onset of neurogenesis were identified based upon WT differential expression between timepoints: early-increasing, genes significantly upregulated between days 20 and 30 (20–30_up^WT) and also days 30 and 60 (30–60_up^WT); early-stable genes, present in 20–30_up^WT and 20–60_up^WT but not 30–60_up^WT; early-transient (20–30_up^WT but not 20–60_up^WT); and late (30–60_up^WT but not 20–30_up^WT). c SZ GWAS enrichment in each transcriptional program, further split into genes that are down-regulated in DLG2^−/− lines at day 30 (e.g., early-stable^−/−) and those that are not (e.g., early-stable^{WT only}). One-sided tests were performed using MAGMA, conditioning on all expressed genes (all^WT+KO); raw and Bonferroni-corrected p values are given (P^GWAS, P^GWAS_corrected), where correction is for the 7 gene-sets tested. d A one-sided, Fisher’s Exact Test was used to identify programs over-represented for the targets of key regulators when compared to all^WT+KO; both raw and Bonferroni-corrected p values are given (P^overlap, P^overlap_corrected), where correction is for the 21 tests performed (3 programs x 7 regulators). All 10 program-regulator enrichments with corrected P < 0.05 were taken forward for genetic analysis. Two-sided tests were performed in MAGMA to investigate whether regulator targets were more highly enriched for SZ association than other genes in that program, conditioning on all^WT+KO and the program as a whole. Raw and Bonferroni-corrected p values are given (P^GWAS, P^GWAS_corrected), where correction is for the 10 sets tested. e SZ GWAS enrichment in GO terms with independent evidence of over-representation amongst early-increasing^−/− genes, two-sided tests were performed using MAGMA, conditioning on all expressed and all early-increasing^−/− genes. Raw and Bonferroni-corrected p values are given (P^GWAS, P^GWAS_corrected), where correction is for the 16 terms tested. Bold indicates tests surviving Bonferroni correction. Source data are provided as a Source Data file.