This study aimed to provide comprehensive diagnostic and candidate analyses in a pediatric rare disease cohort through the Genomic Answers for Kids program. Extensive analyses of 960 families with suspected genetic disorders included short-read exome sequencing and short-read genome sequencing (srGS); PacBio HiFi long-read genome sequencing (HiFi-GS); variant calling for single nucleotide variants (SNV), structural variant (SV), and repeat variants; and machine-learning variant prioritization. Structured phenotypes, prioritized variants, and pedigrees were stored in PhenoTips database, with data sharing through controlled access the database of Genotypes and Phenotypes. Diagnostic rates ranged from 11% in patients with prior negative genetic testing to 34.5% in naive patients. Incorporating SVs from genome sequencing added up to 13% of new diagnoses in previously unsolved cases. HiFi-GS yielded increased discovery rate with >4-fold more rare coding SVs compared with srGS. Variants and genes of unknown significance remain the most common finding (58% of nondiagnostic cases). Computational prioritization is efficient for diagnostic SNVs. Thorough identification of non-SNVs remains challenging and is partly mitigated using HiFi-GS sequencing. Importantly, community research is supported by sharing real-time data to accelerate gene validation and by providing HiFi variant (SNV/SV) resources from >1000 human alleles to facilitate implementation of new sequencing platforms for rare disease diagnoses.

The functional impact of genetic variation has been extensively surveyed, revealing that genetic changes correlated to phenotypes lie mostly in non-coding genomic regions. Studies have linked allele-specific genetic changes to gene expression, DNA methylation, and histone marks but these investigations have only been carried out in a limited set of samples. We describe a large-scale coordinated study of allelic and non-allelic effects on DNA methylation, histone mark deposition, and gene expression, detecting the interrelations between epigenetic and functional features at unprecedented resolution. We use information from whole genome and targeted bisulfite sequencing from 910 samples to perform genotype-dependent analyses of allele-specific methylation (ASM) and non-allelic methylation (mQTL). In addition, we introduce a novel genotype-independent test to detect methylation imbalance between chromosomes. Of the ~2.2 million CpGs tested for ASM, mQTL, and genotype-independent effects, we identify ~32% as being genetically regulated (ASM or mQTL) and ~14% as being putatively epigenetically regulated. We also show that epigenetically driven effects are strongly enriched in repressed regions and near transcription start sites, whereas the genetically regulated CpGs are enriched in enhancers. Known imprinted regions are enriched among epigenetically regulated loci, but we also observe several novel genomic regions (e.g., HOX genes) as being epigenetically regulated. Finally, we use our ASM datasets for functional interpretation of disease-associated loci and show the advantage of utilizing naïve T cells for understanding autoimmune diseases. Our rich catalogue of haploid methylomes across multiple tissues will allow validation of epigenome association studies and exploration of new biological models for allelic exclusion in the human genome.

Genetic variants altering cis-regulation of normal gene expression (cis-eQTLs) have been extensively mapped in human cells and tissues, but the extent by which controlled, environmental perturbation influences cis-eQTLs is unclear. We carried out large-scale induction experiments using primary human bone cells derived from unrelated donors of Swedish origin treated with 18 different stimuli (7 treatments and 2 controls, each assessed at 2 time points). The treatments with the largest impact on the transcriptome, verified on two independent expression arrays, included BMP-2 (t = 2h), dexamethasone (DEX) (t = 24h), and PGE₂ (t = 24h). Using these treatments and control, we performed expression profiling for 18,144 RefSeq transcripts on biological replicates of the complete study cohort of 113 individuals (n_total = 782) and combined it with genome-wide SNP-genotyping data in order to map treatment-specific cis-eQTLs (defined as SNPs located within the gene ±250 kb). We found that 93% of cis-eQTLs at 1% FDR were observed in at least one additional treatment, and in fact, on average, only 1.4% of the cis-eQTLs were considered as treatment-specific at high confidence. The relative invariability of cis-regulation following perturbation was reiterated independently by genome-wide allelic expression tests where only a small proportion of variance could be attributed to treatment. Treatment-specific cis-regulatory effects were, however, 2- to 6-fold more abundant among differently expressed genes upon treatment. We further followed-up and validated the DEX–specific cis-regulation of the MYO6 and TNC loci and found top cis-regulatory variants located 180 kb and 250 kb upstream of the transcription start sites, respectively. Our results suggest that, as opposed to tissue-specificity of cis-eQTLs, the interactions between cellular environment and cis-variants are relatively rare (∼1.5%), but that detection of such specific interactions can be achieved by a combination of functional genomic approaches as described here. Population variation in normal gene expression has been convincingly shown to be under strong genetic control where the main genetic variants are located within close proximity to the gene itself (so called cis-acting). However, the extent to which controlled, environmental stimuli influences cis-regulation of gene expression is unclear. Here, we combine different functional genomic approaches and examine the role of common genetic variants on induced gene expression in a population panel of primary human cells derived from ∼100 unrelated donors treated under multiple conditions. Using these approaches, we find that the interaction between cellular environment and cis-variants are relatively rare, with only a small proportion of the identified genetic variants being specific to treatment. However, although treatment-specific genetic regulation of gene expression seems to be infrequent, we prove its existence by thorough validation of treatment-specific effects of the glucocorticoid-specific regulation of TNC expression. Taken together, these findings indicate that the regulatory landscape within a cell is very stable but, by combining functional genomic tools gene-environmental interactions of clinical importance, can be detected and possibly used as biomarkers in future pharmacogenomic studies.

Personalized antisense oligonucleotides (ASOs) have achieved positive results in the treatment of rare genetic disease¹. As clinical sequencing technologies continue to advance, the ability to identify patients with rare disease harbouring pathogenic genetic variants amenable to this therapeutic strategy will probably improve. Here we describe a scalable platform for generating patient-derived cellular models and demonstrate that these personalized models can be used for preclinical evaluation of patient-specific ASOs. We describe protocols for delivery of ASOs to patient-derived organoid models and confirm reversal of disease-associated phenotypes in cardiac organoids derived from a patient with Duchenne muscular dystrophy (DMD) with a structural deletion in the gene encoding dystrophin (DMD) that is amenable to treatment with existing ASO therapeutics. Furthermore, we designed novel patient-specific ASOs for two additional patients with DMD (siblings) with a deep intronic variant in the DMD gene that gives rise to a novel splice acceptor site, incorporation of a cryptic exon and premature transcript termination. We showed that treatment of patient-derived cardiac organoids with patient-specific ASOs results in restoration of DMD expression and reversal of disease-associated phenotypes. The approach outlined here provides the foundation for an expedited path towards the design and preclinical evaluation of personalized ASO therapeutics for a broad range of rare diseases.

Rare DNA alterations that cause heritable diseases are only partially resolvable by clinical next-generation sequencing due to the difficulty of detecting structural variation (SV) in all genomic contexts. Long-read, high fidelity genome sequencing (HiFi-GS) detects SVs with increased sensitivity and enables assembling personal and graph genomes. We leverage standard reference genomes, public assemblies (n = 94) and a large collection of HiFi-GS data from a rare disease program (Genomic Answers for Kids, GA4K, n = 574 assemblies) to build a graph genome representing a unified SV callset in GA4K, identify common variation and prioritize SVs that are more likely to cause genetic disease (MAF < 0.01). Using graphs, we obtain a higher level of reproducibility than the standard reference approach. We observe over 200,000 SV alleles unique to GA4K, including nearly 1000 rare variants that impact coding sequence. With improved specificity for rare SVs, we isolate 30 candidate SVs in phenotypically prioritized genes, including known disease SVs. We isolate a novel diagnostic SV in KMT2E, demonstrating use of personal assemblies coupled with pangenome graphs for rare disease genomics. The community may interrogate our pangenome with additional assemblies to discover new SVs within the allele frequency spectrum relevant to genetic diseases.