Evaluating gene expression-based approaches for functional assessment of non-coding variants in neurodevelopmental disorders

Non-coding variants associated with neurodevelopmental disorders are difficult to interpret because their regulatory effects are subtle and depend on locus- and stage-specific epigenetic contexts. Transcriptional mis-regulation of nearby genes is the most reliable sign of functional relevance. RNA-seq studies usually detect this via differential gene expression (DEG) analysis, which requires healthy controls and typically more than twofold expression changes. Allele-specific expression (ASE) analysis, may detect subtler effects by directly comparing expression of alleles on the intact and the affected chromosomes within the same sample. We evaluated ASE power using real RNA-seq data, in silico simulations, and a clinical case. We combined transcriptome profiling of induced pluripotent stem cells (iPSCs) and neural progenitor cells (NPCs) with in silico modeling of gene-expression changes to estimate statistical power. We compared two strategies: DEG between variant carriers and non-carriers, and ASE within heterozygous individuals. For clinical validation, we used CRISPR-Cas9 base editing to introduce a synonymous single-nucleotide variant (SNV) adjacent to the candidate regulatory variant in patient-derived NPCs and performed targeted RNA-seq. Simulations and empirical data showed that DEG has higher power for most genes, even with substantial inter-individual or biological variability. ASE surpassed DEG only for genes containing heterozygous SNVs and sequenced at high depth. In a patient carrying a translocation breakpoint near AUTS2 , we generated iPSC and NPC lines with SNV-labeled AUTS2 alleles and devised an inexpensive target-enrichment assay for high-sensitivity ASE profiling, but neither approach detected gene misregulation consistent with the phenotype under current biological models. Our results suggested that DEG and ASE assays are complementary rather than competing strategies. We also demonstrate a practical path to sensitive ASE detection, both in whole-transcriptome and targeted RNA-sequencing, using CRISPR/Cas9-mediated allele labeling. This framework helps researchers to choose cost-effective transcriptomic designs for the functional validation of non-coding variants implicated in neurodevelopmental disorders.