Fig. 5
Alternative splicing, allele-specific expression, eQTL, and network analysis. A Forest plot shows classes of rare variants with significant enrichment (Fisher’s exact test, ** = FDR < 0.05, * = uncorrected p < 0.05) towards genes with alternative splicing in carrier children (n = 9) that are shared with either carrier or noncarrier parents. Forest plot shows log-odds ratios (dots) and 95% confidence intervals (whiskers). Odds ratios, confidence intervals, p values, and Benjamini-Hochberg-corrected FDR values for comparisons with all classes of “second-hit” variants are listed in Additional file 4: Data S2. B Scatter plot shows isoform usage percentage for TADA2A-003 in replicates for individuals in family GL_007. Samples in blue carry a “second-hit” splice-site variant in TADA2A and exhibit a higher frequency of the alternative isoform. C Scatter plot shows allele frequencies for the autism-associated gene CARD11 in carrier child P2C_07, noncarrier parent F2NC_07, and carrier parent M2C_07 in GL_007. Blue circles indicate allele frequency for haplotypes carrying a “second-hit” coding variant disrupting CARD11. D Scatter plot shows z-scores for expression values of ARHGAP39 for all individuals with available genotypes for the gene. Individuals who carry the minor allele for the ARHGAP39 eQTL (blue dots) have higher expression of the gene than the rest of the cohort (orange dots). E Plot shows correlations among the numbers of gene expression alterations in carrier offspring for the 13 trios assessed in our study. Colors and sizes of the circles are proportional to the correlation coefficients between gene expression changes, where blue indicates a positive correlation and red indicates a negative correlation. Asterisks denote significant correlations (FDR < 0.05, Pearson correlation with Benjamini-Hochberg correction). F Boxplot shows the average shortest distances for carrier offspring (n = 13) between pairs of genes with “second-hit” coding variants and genes with identified expression changes in a brain-specific network. Genes with expression changes were more strongly connected to genes with “second-hit” variants in the brain-specific network than the average distances for genes within 100 permuted brain-specific networks per sample (p = 4.88 × 10−4, two-tailed paired Mann-Whitney test). Boxplot indicates median (center line), 25th and 75th percentiles (bounds of box), and minimum and maximum (whiskers)