Whether polygenic risk, monogenic familial hypercholesterolemia (FH), and family history (FamHx) are additively informative for coronary heart disease (CHD) risk prediction across self-identified race/ethnicity (SIRE) groups has not been established. In two diverse cohorts-Electronic Medical Records and Genomics (eMERGE) phase IV (eIV; n = 19,348) and All of Us (AoU; n = 239,645)-we quantified the associations of a polygenic risk score (PRSCHD), pathogenic/likely pathogenic variants in genes associated with FH, and FamHx with CHD and evaluated their incremental value when added to the pooled cohort equations (PCEs). CHD was defined as myocardial infarction, unstable angina, or coronary revascularization. We modeled associations with multivariable logistic regression (prevalent CHD in eIV) and Cox proportional hazards (incident CHD in AoU) and characterized predictive performance with the c-statistic and reclassification and decision-curve net benefits across actionable 10-year risk thresholds. The effects of PRSCHD and FamHx were independent and additive in both cohorts and consistent across White, Black, and Latino SIRE groups. In eIV, adding PRSCHD and FamHx to the PCE increased the c-statistic for prevalent CHD from 0.719 to 0.753 (p-diff = 9.1 × 10[-3]) and reclassified 18.8% of participants at the 7.5% 10-year threshold, yielding approximately 4 additional true-positive CHD identifications per 1,000 screened. Net benefit gains were observed between the 7.5% and 10% thresholds across all three SIRE groups. In conclusion, PRSCHD and FamHx were independently and additively associated with CHD across major SIRE groups in two diverse cohorts in the United States (US), motivating the addition of these factors to clinical risk algorithms.

Early postzygotic mutations (PZMs) that arise after fertilization but prior to primordial germ cell specification may be present in both somatic and germ cells, causing mosaicism in a parent and constitutive inheritance in their offspring. In clinical family-trio whole-genome sequencing (WGS), such variants are systematically missed because their sub-heterozygous variant allele fraction (VAF) prevents heterozygous calling in the parent, while residual parental allele support disqualifies the variant as a candidate germline de novo mutation (DNM) in the child. Here, we developed a bioinformatic approach to ascertain parental PZMs from unfiltered DNM candidates in standard-depth (∼30×) trio WGS and applied it to 12,015 trios from the Genomics England 100,000 Genomes Project. We identified 1,015 high-confidence early autosomal parental PZMs, a large single-source catalog of this mutation class. These exhibited a monomodal VAF distribution centered around 5% in parental blood, consistent with empirically characterized ascertainment boundaries imposed by standard-depth sequencing and germline variant calling. PZMs showed no parental age or sex bias and displayed a mutational spectrum distinct from that of DNMs, with enrichment for C>A and T>A substitutions and depletion of T>C. Mutational signature analysis revealed that both mutation types are shaped by clock-like signatures SBS1 and SBS5 in similar proportions, suggesting that spectral differences reflect shifts within shared mutagenic processes. Exploratory genomic distribution analysis revealed a negative PZM association with GC content, in contrast to the positive association for DNMs. Among these, we found variants in DYNC1H1 and WT1 with potential clinical relevance that were missed by routine diagnostic pipelines.

Multiplexed assays of variant effects (MAVEs) represent a powerful approach to providing functional information for variants at scale. To harness the full utility of these systems, assay readout must be translated into a language that is accommodating to the clinical genomics community. We previously performed a MAVE to characterize variants in JAG1, the primary cause of the autosomal-dominant, multisystemic disease Alagille syndrome. Using this existing dataset, we calculated log-likelihood ratios of pathogenicity for each variant score, allowing for direct translation of variant data into recognized evidence weights utilized by the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) in clinical variant classification. Calibration resulted in improved separation of known benign and pathogenic variants and increased the classification rate of abnormal missense variants from 486 to 610, providing clear binning for strong (n = 1), moderate (n = 340), and supporting (n = 269) evidence toward pathogenicity. Retrospective application of this evidence to a cohort of 29 individuals with a JAG1 variant of uncertain significance (VUS) identified from clinical diagnostic sequencing yielded nine (31%) variants with abnormal data meeting supporting (n = 3) and moderate (n = 6) weight for pathogenicity, of which six (21%) were upgraded to likely pathogenic or pathogenic. Calibration of MAVE data to comply with the ACMG/AMP variant classification framework improves the diagnostic yield for JAG1 variants for Alagille syndrome. Moreover, our results support the broader application of these models to additional MAVEs, suggesting that they are likely to strongly impact variant classification across genes associated with disease.

Cohesin orchestrates gene expression via three-dimensional chromosome folding. Genes encoding cohesin and cohesin loaders have been associated with Mendelian disorders, whereas genes encoding cohesin release factors, including WAPL and its binding partners PDS5A and PDS5B, have not. We explored the relevance of cohesin release factors in Mendelian disease by phenotyping individuals with heterozygous predicted damaging variants in WAPL (n = 27), PDS5A (n = 8), and PDS5B (n = 8), by modeling WAPL deficiency in human cells and mice, and by aggregating disease association statistics from consortia studies. We identified a WAPL-related disorder featuring developmental delay, intellectual disability, and risk of other developmental anomalies. Similarities between individuals with damaging WAPL variants and those with large, recurrent 10q22.3q23.2 (10q) deletions encompassing WAPL nominate WAPL as a driver gene within this genomic disorder region. While individuals with PDS5A or PDS5B variants exhibited features of developmental disorders, neither cohort-based statistics nor subject phenotyping associated these genes with specific phenotypes. We used CRISPR to generate truncating variants in WAPL and 10q deletion or duplication in human induced pluripotent stem cells (iPSCs) and induced neurons. Transcriptomics identified significant overlap between WAPL haploinsufficiency and 10q deletion differentially expressed genes. Mice with 50% Wapl expression exhibited mild deficits of growth and learning/memory, whereas those with 25% residual Wapl displayed birth defects and postnatal lethality, revealing a dosage liability threshold below the level of heterozygosity. In summary, we delineated a genetic condition caused by cohesin release factor deficiency, nominated WAPL as a driver gene within a genomic disorder region, and further illuminated dosage sensitivity of human cohesin.

Hereditary hypotrichosis comprises a group of nonsyndromic hair growth disorders for which molecularly characterized forms have been predominantly attributed to autosomal inheritance. Here, we identified three unrelated families with X-linked hypotrichosis simplex (XLHS), characterized by normal hair at birth followed by progressive scalp hair sparsity that emerges in childhood or adolescence. Affected males showed more severe and diffuse involvement, whereas affected females exhibited milder, patchy thinning, consistent with X-linked inheritance. Genomic analyses revealed partially overlapping duplications at Xq13.3 in all affected individuals, defining a shared critical interval encompassing TAB3, FTHL17, and part of DMD. Among these genes, FTHL17 is mainly expressed in the testicular system, and DMD is classically associated with muscular dystrophy, whereas neither FTHL17 nor DMD showed detectable differences in protein abundance between affected individuals and healthy control subjects in scalp tissue. In contrast, TAB3 showed increased accumulation in scalp tissue from affected individuals compared with healthy control subjects, supporting TAB3 as the most likely dosage-sensitive candidate gene. Functionally, both an inducible keratinocyte-specific Tab3-overexpressing mouse model and TAB3-overexpressing HaCaT keratinocyte models demonstrated impaired TAK1-dependent canonical NF-κB signaling. In mice, Tab3 overexpression recapitulated the human hypotrichosis phenotype and induced early pathological changes on postnatal day 31, including hyperkeratosis, thickening of the stratum corneum, and inflammatory cell infiltration. Anti-inflammatory treatment improved hair density in mice, and clinical observations in affected individuals suggested potential benefit. Together, these findings define a previously unrecognized XLHS, implicate an increase in TAB3 dosage as its genetic driver, and identify dysregulated TAK1-NF-κB signaling as a candidate pathogenic mechanism.

CDK20, also known as CCRK, is a cyclin-dependent kinase that regulates cell growth, primary cilium structure, and Sonic Hedgehog signaling. CDK20 has not yet been associated with a known human disorder. In this report, we present a cohort of seven individuals from five unrelated families with bi-allelic variants in CDK20 and an overlapping phenotype of ventriculomegaly or hydrocephalus, midline brain anomalies, abnormal nose, midline cleft lip and/or palate, cryptophthalmos or anophthalmia, postaxial polydactyly, and a sandal toe gap. Immunoblot analysis of fibroblasts derived from two affected fetuses with a homozygous CDK20 c.687+6T>C (p.?) variant demonstrated reduced CDK20 levels. Fibroblasts derived from these affected fetuses were also significantly deficient in cilium formation and function, with abnormal cilium morphology and significantly decreased Hedgehog responsiveness. Our findings support that bi-allelic loss-of-function variants in CDK20 cause a severe ciliopathy with midline brain and facial anomalies.

Genome-wide association studies (GWASs) have identified at least 17 genetic variants associated with gastric cancer risk. However, the underlying genetic regulatory mechanisms remain poorly understood. We systematically performed genome-wide analyses of expression quantitative trait loci (eQTLs), splicing QTLs (sQTLs), and alternative polyadenylation QTLs (apaQTLs) using gastric biopsy tissues from 262 Chinese individuals, and we integrated these data with GWAS results from 10,254 gastric cancer cases and 10,914 controls. We identified 4,636 eGenes, 1,422 sGenes, and 511 apaGenes, with 55.3% of sGenes and 49.5% of apaGenes being distinct from eGenes. Notably, over half of the variants associated with shared genes were unique to a specific QTL type, with these variants characterized by a wide physical separation (median distance = 25.01 kb) and weak linkage disequilibrium (median r2 = 0.42). Gastric cancer risk variants showed significant enrichment across all three QTL categories. Integrative transcriptome-wide association studies (TWASs) across expression, splicing, and alternative polyadenylation identified 34 genes at false discovery rate (FDR) <0.05; PSCA was implicated across all three regulatory layers. The candidates DIP2B and RMC1 were experimentally validated to promote gastric tumorigenesis both in vitro and in vivo. Functional analyses revealed that the rs11264361-T allele reduced FDPS exon 9 skipping, leading to increased expression of the full-length transcript, while the apaVariant rs7445 regulated UBE2L3 polyadenylation via ATXN2 binding, contributing to gastric cancer risk. Collectively, these results provide a comprehensive QTL resource in gastric tissues and demonstrate distinct regulatory mechanisms of eQTLs, sQTLs, and apaQTLs in gastric cancer development.

Accurately characterizing genetic ancestry is critical for ensuring reproducibility and fairness in genomic studies and downstream health research. This study aims to address the prediction of ancestry from genetic data using deep learning, with a focus on generalizability across datasets with diverse populations and on explainability to improve model transparency. We adapt the Diet Network, a deep-learning architecture proven to be effective in handling high-dimensional data, to learn population ancestry from single-nucleotide polymorphism (SNP) data using the populational Thousand Genomes Project dataset. Our results highlight the model's ability to generalize to diverse populations in the CARTaGENE, Montreal Heart Institute, and All of Us biobanks and that predictions remain robust to high levels of missing SNPs. We show that, despite the lack of North African populations in the training dataset, the model learns latent representations that reflect meaningful population structure for North African individuals in the biobanks. To improve model transparency, we apply Saliency Maps, DeepLift, GradientShap, and Integrated Gradients attribution techniques and evaluate their performance in identifying SNPs leveraged by the model. Using DeepLift, we show that the model's predictions are driven by population-specific signals consistent with those identified by traditional population-genetics metrics. This work presents a generalizable and interpretable deep-learning framework for genetic-ancestry inference in large-scale biobanks with genetic data. By enabling more widespread genomic ancestry characterization in these cohorts, this study contributes practical tools for integrating genetic data into downstream biomedical applications, supporting more inclusive and equitable healthcare solutions.

Transcriptomic diversity across individuals arises from multiple modes of RNA regulation-including precursor messenger RNA (pre-mRNA) expression, splicing, degradation, and other processes-and has been widely leveraged to map molecular quantitative trait loci (xQTLs) and interpret genome-wide association study (GWAS) signals. We recently developed a multimodal framework called Pantry that can extend discovery beyond total expression by integrating multiple transcriptomic modalities. However, Pantry and similar tools remain limited by their reliance on complete gene annotations and the statistical complexity of jointly analyzing correlated modalities. Here, we present LaDDR (latent data-driven RNA phenotyping), a mechanism-agnostic framework that generates orthogonal, latent coverage features per gene, enabling xQTL discovery and GWAS integration without requiring complete gene annotations. Applied to the Genotype-Tissue Expression (GTEx) Project, LaDDR identified an average of 95% more independent xQTLs per tissue than the six transcriptional regulation modes implemented in Pantry ("knowledge-driven"). Residualizing known modalities prior to LaDDR and combining with knowledge-driven phenotypes increased discovery by an additional 41% per tissue on average while retaining the interpretability of knowledge-driven signals. In a transcriptome-wide association study (TWAS) of 114 complex traits, the use of LaDDR-derived phenotypes uncovered an average of 11,796 unique gene-trait pairs per tissue versus 8,630 from knowledge-driven phenotypes. The newly captured genetic signals exhibit functional and colocalization qualities consistent with known mechanisms, suggesting that LaDDR broadens the detectable landscape of trait-relevant transcriptomic regulation by efficiently recovering regulatory variation missed by current pipelines.

Locus 9p21.3 was the first genome-wide significant locus for coronary artery disease (CAD) and replicates across multiple non-African populations, yet it is absent in African-ancestry cohorts. We analyzed multi-ancestry data from European, East Asian, South Asian, Middle Eastern, African, and admixed American groups. Ancestry was inferred using both global and local-ancestry inference (LAI) approaches. Within each ancestry, we performed CAD associations at 9p21.3 with common and rare variants, conditional and haplotype analyses, and pleiotropy assessments. CAD associations at 9p21.3 were robust in European, East Asian, Middle Eastern, and admixed American ancestry groups, nominal in South Asians, and absent in Africans despite similar allele frequencies for lead-associated SNPs. LAI-stratified analyses showed strong signals in African heterozygotes carrying 9p21.3 European chromosomes but not in homozygous Africans. The association is due to highly common variants; rare variants did not explain the locus signal. Linkage disequilibrium (LD) blocks were extended in non-Africans but fragmented into smaller blocks in Africans. Fine-mapping identified multiple independent signals, including an East Asian-specific haplotype absent in Europeans and Africans. Hamming-distance analyses revealed that risk alleles are dispersed across multiple haplotypes in homozygous-African groups. Phenome-wide association study (PheWAS) mirrored these ancestry-specific patterns across metabolic traits. The non-replication of 9p21.3 in African ancestry is not due to absent risk alleles, reduced power, or poorer imputation of risk alleles due to weaker LD; rather, it reflects greater haplotype diversity that disperses risk alleles across configurations, attenuating case-control contrast. These findings illustrate how ancestry-specific LD and multi-causal haplotype architecture modulate association detectability.

The role of heritable genetic variation in hematologic traits in adults is well established, yet the genetic architecture of neonatal blood cell traits is unknown. Leveraging flow cytometry profiling in cord blood samples from 382 Hispanic newborns, we conducted genome-wide association studies (GWASs) of 24 blood cell phenotypes to assess the impact of genetic and birth-related characteristics. We identified six genome-wide significant loci, including a signal at chromosome 2p11.2 spanning CD8A and CD8B that was associated with CD4[+]CD8[+] double-positive T cells (rs35505884-C: p = 6.83 × 10[-26]). The lead variant showed strongly differentiated allele frequencies between European (22.1%) and admixed American (11.8%) populations and was associated with global (p = 2.51 × 10[-6]) and local (p = 2.01 × 10[-8]) Indigenous American ancestry. Colocalization with single-cell expression quantitative trait loci (eQTLs) revealed convergence of GWAS effects with regulation of CD8A expression in natural killer cells and CD4[+] effector memory T cells with posterior probability (PPH4) >0.85. A locus at 6p21.33 was associated with CD4[-]CD8[+] cytotoxic T cells (rs2853973-C, p = 3.90 × 10[-8]) and colocalized with MICA-AS1 expression across multiple immune cell types (PPH4 > 0.85). Additional nominal genome-wide significant loci were detected for total CD45[+] lymphocytes, CD19[+]CD56[+]CD16[+] natural killer cells, and CD4[+]CD8[-] helper T cells. C-section delivery and biological sex significantly altered blood cell profiles, with the largest magnitudes of association observed between C-section and total CD3[+] T cells (β = 0.39) and CD4[+] helper T cells (β = 0.39), and between female sex and CD4[+]CD8[-] (β = 0.39) and CD4[-]CD8[+] T cells (β = -0.36). Our study demonstrates that blood cell composition at birth is under strong genetic regulation in Hispanic newborns, highlighting the value of studying complex traits in early life.

Mendelian randomization has emerged as a transformative approach for inferring causal relationships between risk factors and disease outcomes. However, applying Mendelian randomization to disease progression-a critical step in validating therapeutic targets-is hampered by index event bias. This form of selection bias occurs because analyses of disease progression are necessarily restricted to individuals who have already experienced the disease event. Here, we present a comprehensive evaluation of statistical methods designed to mitigate index event bias, including inverse-probability weighting, Heckman's method, the Slope-Hunter method, and multivariable methods. We compare their performance in simulations and applied examples. Inverse-probability weighting methods reduce bias but require individual-level data and will only fully eliminate bias when the disease event model is correctly specified. The Slope-Hunter method performed poorly in all simulation scenarios, even when its assumptions were fully satisfied. Multivariable methods worked best when including genetic variants that affect the incident disease event. However, if these genetic variants also affect disease progression directly, then the analysis will suffer from pleiotropy. Hence, if the same biological mechanisms affect disease incidence and progression, then multivariable methods will have little utility. In such a case, however, analyses of disease progression are less critical, as conclusions reached from analyses of disease incidence are likely to hold for disease progression. Our findings indicate that no single method is a universal solution to provide reliable results for the investigation of disease progression. Instead, we propose a strategic framework for method selection based on data availability and biological context.

Genomic medicine promises to leverage research on human genetic variation to improve health. As technology improves, our increasing ability to determine an individual's genomic sequence must be matched with increased understanding of underlying mechanisms of disease and how phenotypes arise because of genetic variation. Knowledge about the etiology of rare monogenic conditions can be leveraged for medical and public health purposes through diagnostic testing and predictive screening. However, policy decisions about the use of genetic and genomic analysis in a predictive setting must consider the penetrance and expressivity of each monogenic condition and how these factors influence the timing of genomic screening and its potential benefits and harms. This means that high-quality data are required on the nature of genetic diseases and how they impact human health across populations. The objective of this paper is to address the inherent challenges in determining the onset of symptoms for rare monogenic conditions and underscore the importance of curating natural history data. By elucidating the natural progression of rare monogenic conditions, we can optimize screening strategies and guide clinical decision-making. Detailed information about natural history is critical to understanding the optimal timing of population screening and subsequent intervention for any given condition.