Genetics in Medicine

Genomic sequencing of healthy newborns to screen for medically important genetic information has long been anticipated but data around downstream medical consequences are lacking. Among 159 infants randomized to the sequencing arm in the BabySeq Project, an unanticipated monogenic disease risk (uMDR) was discovered in 18 (11.3%). We assessed uMDR actionability by visualizing scores from a modified ClinGen Actionability SemiQuantitative Metric and tracked medical outcomes in these infants for 3-5 years. All uMDRs scored as highly actionable (mean 9, range: 7-11 on a 0-12 scale) and had readily available clinical interventions. In 4 cases, uMDRs revealed unsuspected genetic etiologies for existing phenotypes, and in the remaining 14 cases provided risk stratification for future surveillance. In 8 cases, uMDRs prompted screening for multiple at-risk family members. These results suggest that actionable uMDRs are more common than previously thought and support ongoing efforts to evaluate population-based newborn genomic screening.

IntroductionVariant-level functional data is a core component of clinical variant classification and can aid in reinterpreting variants of uncertain significance (VUS). However, the usage of functional data by genetics professionals is currently unknown. MethodsAn online survey was developed and distributed in spring of 2024 to individuals actively engaged in variant interpretation. Quantitative and qualitative methods were used to assess responses. Results190 eligible individuals responded, with 93% reporting interpreting 26 or more variants per year. The median respondent reported 11-20 years of experience. The most common professional roles were laboratory medical geneticists (23%) and variant review scientists (23%). 77% reported using functional data for variant interpretation in a clinical setting and overall respondents felt confident in assessing functional data. However, 67% indicated that functional data for variants of interest was rarely or never available, and 91% considered insufficient quality metrics or confidence in the accuracy of data as barriers to its use. 94% of respondents noted that better access to primary functional data and standardized interpretation of functional data would improve usage. Respondents also indicated that handling conflicting functional data is a common challenge in variant interpretation that is not performed in a systematic manner across institutions. DiscussionThe results from this survey showed a demand for a comprehensive database with reliable quality metrics to support use of functional evidence in clinical variant interpretation. The results also highlight a need for guidelines regarding how putatively conflicting functional data should be used for variant classification.

IntroductionThe Clinical Genome Resource (ClinGen) Von Hippel-Lindau (VHL) Variant Curation Expert Panel (VCEP) has created variant classification specifications tailored to the VHL gene, including phenotype-driven and evidence-based criteria, somatic and germline mutational hotspots, functional and in-silico data. Materials and MethodsUsing the American College of Medical Genetics and Genomics (ACMG) guidance and the ClinGen Sequence Variant Interpretation (SVI) recommendations, the VCEP made substantial modifications to eight evidence codes (PVS1, PS3, PS4, PM1, BS2, BS3, BS4, BP5), while 14 had minor or no changes and 6 were not used (PM3, PP2, BP1, PP4, PP5/BP6). The VHL VCEP applied two literature sets of over >428 papers in Clinical Interpretations of Variants in Cancer (CIViC) and >8700 structured annotations using Hypothesis. ResultsFrom 31 pilot variants, 15 remained pathogenic/likely pathogenic, 9 resolved to benign through the stand-alone benign evidence code and 7 variants with initial uncertain classifications, with many lacking additional literature, remained uncertain. ConclusionThe versioned VHL VCEP specifications are publicly available in the ClinGen Criteria Specifications Registry and will enhance the transparency and consistency of variant classifications for this highly sequenced hereditary cancer gene.

Yield of reported results from genetic testing provides a proximal measure of clinical usefulness. While ACMG/AMP guidelines provide representations of uncertainty for individual genetic variant classification, additional factors are considered when determining whether results explain a patient's presentation. To standardize cross-consortium analysis, a working group of the Clinical Sequencing Evidence-Generating Research (CSER2) consortium iteratively identified factors used when contextualizing variant-level results to case-level interpretation (i.e., interpretation of an individual's genetic data with respect to the indication for testing). Sites independently categorized results; complex cases were discussed collaboratively, leading to revision of classification categories. Our metric incorporates factors beyond classification of reported variants. Analogous to variant-level results, "Definitive Positive" and "Probable Positive" represent certainty that results may be clinically explanatory. The category "Inconclusive" applies when results may or may not fully explain the patient presentation, with subdivision into multiple (non-exclusive) subcategories. Cases falling outside all of the other categories are considered "Negative". The overall diagnostic yield by this metric and use of categories for inconclusive results varied by CSER project, in part paralleling study design differences. This case-level categorization provides a meaningful assessment of diagnostic yield, and for inconclusive cases identifies potentially resolvable factors for case resolution.

Purpose: Genetic variant reclassification is increasingly common in clinical genomics, yet limited data describe how patients experience re-contact and variant reclassification in routine clinical care. Methods: We conducted semi-structured qualitative interviews with 20 adult patients who received a variant reclassification following routine clinical genetic testing. Interviews explored emotional responses, communication experiences, and perceived value of genetic testing. Data were analyzed using Template Analysis, a form of thematic analysis. Results: Three overarching themes were identified. Participants identified a need for improved communication of reclassified results, particularly with respect to timing, modality, and contextualization (Theme 1). Experiences with reclassification also shaped perceptions of the value of genetic testing, with most participants viewing testing as worthwhile despite its evolving nature (Theme 2). Finally, many participants interpreted reclassification as evidence of personalized and ongoing care, reinforcing trust in genetic testing and biomedical research (Theme 3). Participants generally preferred to be informed of reclassified results regardless of reclassification type, although the direction of reclassification influenced emotional responses and preferred modes of communication. Downgrades from variants of uncertain significance to benign or likely benign were widely viewed as meaningful by participants. Conclusion: Variant reclassification was experienced as a signal of personalized, ongoing care. Timely, contextualized, patient-centered re-contact practices may reduce uncertainty, strengthen trust, and help patients not feel forgotten.

Precision medicine increasingly relies on genetic testing for individualized care, yet the practice patterns of genetic evaluation in adults remain under-characterized and few guidelines exist to inform appropriate testing. We analyzed eight years of electronic health record (EHR)-linked data from a high-volume adult genetics clinic (9,867 visits) within a major academic health system to define referral patterns, test utilization, and genetic testing outcomes across indications and demographic groups. Genetic testing was ordered for 52% of all new patients, with significant indication-specific variation in ordering propensity. Overall, 24% of all tests returned a diagnostic result. Diagnostic yield differed markedly by testing modality: whole-exome sequencing yielded diagnoses in 40% of patients, whereas next-generation sequencing panels, the most frequently ordered testing type, yielded diagnoses in only 16%. Although diagnostic yield declined with increasing patient age it remained above 17% in every age stratum, with the yield of exome/genome sequencing remaining above 30% across all age groups. Outcomes varied significantly by referral indication, with some yielding high rates of pathogenic findings, and others more often yielding negative or uncertain results. Operational factors materially shaped practice: test type and laboratory utilization shifted significantly over time in step with payer and testing lab policy changes. Together, these practice-based data establish adult genetics as a high-yield and important clinical domain. Our findings provide actionable evidence to inform age- and indication-aware triage, guide workforce planning, promote adult-focused practice guidelines, and advance the integration of genomic medicine into routine adult patient care.

As population DNA sequencing becomes more common, genomic-first approaches are increasingly used to identify individuals with possible rare genetic disorders. To accurately estimate prevalence and penetrance, these studies often confirm manifestation of the disorder using electronic health records (EHRs). Multiple strategies exist to search the EHR for diagnoses of rare disorders, however, each has its limitations. We have developed a portable, ensemble tool, DxFit, that mines EHR data (ICD codes and structured diagnosis descriptions from billing code and problem list tables) for a diagnosis consistent with a given rare genetic disorder. DxFit combines evidence across four strategies: (1) gene name searches in diagnosis descriptions and notes, (2) ICD conversion to Mondo rare disorder ontology to find exact and nearby matches, (3) word embedding similarity searches, and (4) Jaccard similarity matches. DxFit prioritizes the match type and outputs the most confident match for each participant-disorder pair. On a cohort of 350 participants with a known positive result from diagnostic genetic testing for developmental disorders, DxFit had a sensitivity of 88.7% and specificity of 86.2% using default parameters. Adjusting the linguistic scoring thresholds from 0.8 to 0.7 and allowing for synonymous matches yielded a sensitivity of 92.7% and specificity of 84.5%. Partitioning EHR evidence into windows before and after genetic testing demonstrates, as expected, that the overall DxFit rates increase after testing and the match types become more confident. DxFit is available to the public and has extensive customization options to support a wide range of uses. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=187 HEIGHT=200 SRC="FIGDIR/small/720629v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@d71d00org.highwire.dtl.DTLVardef@b11a9eorg.highwire.dtl.DTLVardef@14a9304org.highwire.dtl.DTLVardef@fa23aa_HPS_FORMAT_FIGEXP M_FIG C_FIG

PurposeDiagnostic yield from exome and genome sequencing varies widely across studies. It remains unclear how much of this variation reflects patient-level factors (e.g., sex, clinical features, race/ethnicity, genetic ancestry) versus site-level practices such as sequencing modality or variant interpretation workflows. We aimed to quantify the contributions of these factors to diagnostic outcomes across five U.S. clinical sequencing sites. MethodsWe performed a cross-sectional analysis of 3,008 prenatal, neonatal, and pediatric cases from the NHGRI Clinical Sequencing Evidence-Generating Research (CSER) consortium (2017-2023). Clinical indications spanned neurodevelopmental, neurological, immunological, metabolic, craniofacial, skeletal, cardiac, prenatal, and oncologic presentations. Genetic ancestry was inferred from sequencing data, and variants were interpreted using ACMG/AMP guidelines to classify DNA-based diagnoses. Generalized linear mixed models were used to estimate associations between diagnostic yield and fixed effects (sex, prenatal status, isolated cancer, number of clinical indications, sequencing modality, race/ethnicity, and genetic ancestry), while modeling study site as a random effect to quantify between-site variation. ResultsThe overall diagnostic yield was 19.0%. Multiple clinical indications (OR=1.47, 95% CI 1.20-1.80, p<0.001) were associated with higher diagnostic yield, and male sex (OR=0.80, 95% CI 0.66-0.96, p=0.017) and prenatal status (OR=0.63, 95% CI 0.44-0.90, p=0.012) were associated with lower yield. Sequencing modality, race/ethnicity, genetic ancestry, and isolated cancer were not statistically significantly associated with diagnostic outcomes.. A model without fixed effects attributed [~]10% of variance in diagnostic yield to between-site differences. After adjusting for covariates, site-level variance decreased to 5.7%, indicating consistent variation across sites not explained by measured patient factors. ConclusionAcross five sites, patient-level clinical features influenced diagnostic yield, but substantial site-level variation remained even after adjustment. Differences in variant interpretation, or case-classification practices may contribute to this residual variability. Further efforts to increase consistency in exome- and genome-sequencing diagnostic workflows may help reduce inter-site differences.

The sequencing of the first human genome led to expectations of the widespread use of genetics in medicine. However, assessing the true impact of genetic testing on clinical practice is challenging due to the lack of integration in the electronic health record (EHR). We extracted clinical genetic tests from the EHRs of over 1.8 million patients seen at Vanderbilt University Medical Center from 2002 to 2022, using both automated and manual methods. Using these data, we quantified the extent of clinical genetic testing in healthcare and described how testing patterns have changed over time, including utilization rate, test comprehensiveness, diagnoses made, and the number of variants of uncertain significance (VUS) returned. We also assessed genetic testing rates across medical specialties and introduce a measure - the genetic attributed fraction (GAF) - to compute the proportion of observed phenotypes attributable to a genetic diagnosis. We identified 104,392 tests, 32% of which were only reported in unstructured text, and 19,032 molecularly confirmed diagnoses or risk factors. The proportion of patients genetic testing recorded in their EHRs from 1.0% in 2002 to 6.1% in 2022, and testing became more comprehensive with the growing use of multigene panels. This corresponded with a substantial increase in the variety of diseases diagnosed with genetic testing, from 51 unique diseases in 2002 to 509 in 2022, alongside a growing number of VUS. The phenome-wide GAF for 6,505,620 diagnoses made in 2022 was 0.46%, with 74 phenotypes having a GAF greater than 5%, including pancreatic insufficiency (67%), chorea (64%), atrial septal defect (24%), Microcephaly (17%), paraganglioma (17%), and ovarian cancer (6.8%). Our study provides a comprehensive quantification of the increasing role of genetic testing at a major academic medical institution. These results demonstrate the now pervasive use of genetic testing across diverse medical contexts and its growing utility in explaining observed medical phenome.

Various professional organizations recommend screening prospective parents for autosomal recessive (AR) and X-linked (XL) conditions, which is reflected in commercial screening panels. There is merit to developing a distinct reproductive gene-list and analytic framework inclusive of genes based on available perinatal intervention, defined as possible prenatal intervention (including investigational) for the fetus or necessary early initiation of approved postnatal treatments. We evaluated a reproductive genetic screening framework that incorporates perinatal actionability across AR, XL, and selected autosomal dominant (AD) genes. Using a curated list of genetic conditions with perinatal intervention, we evaluated five subset gene lists to determine the individual-level number-needed-to-screen (NNS) to identify one individual with at least one qualifying heterozygous variant, defined as a heterozygous pathogenic or likely pathogenic (P/LP) variant in a gene on the specified list. To conduct NNS analyses, we sourced carrier frequency and allele frequency data for each gene and their respective ClinVar-curated high-confidence (>=2 star) P/LP variants, from two population databases -- gnomAD v4.1 and All of Us (AoU) v8. The analyses produced an individual-level NNS of 3.20 (CI: 3.193, 3.212) using gnomAD and 3.62 (CI: 3.606, 3.640) using AoU. These estimates do not represent couple-level reproductive risk, affected-pregnancy yield, clinical diagnostic yield, or validation of a clinical screening test. These findings support further evaluation of a perinatal-actionability framework, with clinical value dependent on which genes drive yield, and whether the relevant gene, variant, mechanism, and phenotype combinations are actionable in a reproductive or perinatal context for both the pregnant woman and her future offspring.

Purpose Children with suspected rare genetic disorders often remain undiagnosed after exome, gene panel, or chromosomal microarray testing. We sought to define recurrent missed-diagnosis mechanisms and evaluate an integrated clinical-genomic reanalysis framework. Methods We studied 75 participants with rare or striking pediatric-onset phenotypes who remained undiagnosed after prior genetic testing. Trio genome sequencing evaluated SNVs, indels and copy-number variants. RNA sequencing was performed in 63 participants to assess aberrant expression and splicing, with targeted follow-up assays used selectively for prioritized variants. Clinician-genome scientist review integrated longitudinal phenotyping, physician input, and inheritance. Unresolved cases underwent research-level candidate prioritization. Results Clinically reportable findings were returned in 9, including two with multilocus findings: 5 explained the primary phenotype and 4, while not fully explanatory, were clinically relevant and informing surveillance, management, or family follow-up. Previously missed findings reflected technical limitations, including deep intronic splice-altering variants supported by transcriptome analysis and a genomic deletion, and interpretive limitations related to incomplete phenotype integration or narrowly scoped prior testing. Research-level analysis prioritized six candidate genes, including biallelic ACHE variants with reduced enzymatic activity. Conclusion Integrated clinical-genomic reanalysis framework identified diagnostic gaps, recovered care-informing findings, and prioritized candidate genes for rare disease discovery.

The clinical utility of genomic testing is constrained by variants of uncertain significance (VUS), which complicate diagnostic interpretation and patient management. The ACMG/AMP PS4 criterion, "prevalence in affected individuals statistically increased compared to controls," offers strong evidence for pathogenicity but is often challenging to apply due to the limited availability of robust, matched case-control genomic and phenotypic data. Further, there are currently no options available to score evidence from case-control studies towards benignity. We propose and validate a new code, RWE (real world evidence), by integrating de-identified, longitudinal clinical data with variant carriers and non-carriers identified from exome or genome sequence data across three large-scale clinicogenomic datasets: the Helix Research Network (HRN) dataset, UK Biobank (UKB) and All of US (AoU). Phenotypes for established gene-level disease associations were compiled from the longitudinal medical records of the individuals, enabling rigorous variant-specific case-control analyses from population data. This RWE approach was systematically applied to all variants, including previously identified VUS in clinically relevant genes, powering our VUS Early Surveillance Platform. Across 20 hereditary cancer and cardiovascular genes, the application of RWE provided sufficient evidence to reclassify a VUS in 32% of VUS carriers-99.7% to B/LB and 0.3% to P/LP-directly resolving their ambiguous status. This reclassification rate varied by gene, ranging from 0.7% for BRCA2 up to 50% for LDLR. The systematic integration of Real-World Evidence from large-scale clinicogenomic datasets into the ACMG/AMP scoring rubric through our newly developed and statistically robust RWE category is a significant improvement to variant interpretation that is projected to resolve over 50% of VUS carriers once longitudinal clinico-genomic databases are available for ~3M individuals. This approach markedly reduces the burden of Variants of Uncertain Significance, provides more definitive diagnoses for a substantial proportion of previously unresolved cases, and ultimately increases the clinical utility and adoption of genomic testing, representing a critical advancement for precision medicine.

Neurodevelopmental disorders (NDDs) are clinically and genetically heterogeneous, and evidence linking additional genes to disease evolves rapidly. We developed SysNDD, an open-access, expert-curated resource that represents distinct gene-inheritance-disease associations as entities with standardized confidence, phenotype, and variant annotations. Each entity retains its curation history, so its evidence category can change as evidence accumulates. Currently, SysNDD contains 4,275 entities curated from 4,932 publications. Of these, 3,758 entities involving 3,271 genes describe NDDs; the remainder record non-NDD disorders, nearly all linked to NDD-associated genes, included for completeness. Regarding NDD evidence categories, 1,838 genes are Definitive, 179 Moderate, 1,248 Limited, and 6 Refuted. The entity model captures inheritance-specific disorders, including 220 genes with both autosomal dominant and autosomal recessive NDD associations. Comparison with seven other NDD gene resources showed complementary coverage, and evidence categories largely agreed with those of ClinGen, Gene2Phenotype, and PanelApp for shared gene-disease pairs. NDD genes formed distinct molecular modules, whereas associated disorders formed broad, overlapping phenotype groups that corresponded to those modules only selectively. SysNDD provides a continuously maintained, programmatically accessible framework for gene-disease evidence assessment, inheritance-aware interpretation, and systematic NDD research.

PurposeSouthSeq, a translational research study to perform genome sequencing (GS) for infants with symptoms suggestive of a genetic disorder, was conducted in NICUs in the Southeastern US. Recruitment targeted racial/ethnic minorities and rural, medically underserved areas that are historically under-represented in genomic medicine research. MethodsGS and analysis were performed for 367 newborns to detect disease-causal genetic variation concurrent with standard of care evaluation and testing. ResultsDefinitive diagnostic (DD) or likely diagnostic (LD) genetic findings were identified in 30% of newborns and 14% harbored an uncertain result. Only 39% of DD/LD findings were identified via concurrent standard of care suggesting that GS testing is better for obtaining early genetic diagnosis. We also identified phenotypes that correlate with the likelihood of receiving a DD/LD finding, such as craniofacial, ophthalmologic, auditory, skin, and hair abnormalities. We did not observe any differences in diagnostic rates between racial/ethnic groups. ConclusionWe describe one of the largest to-date GS cohorts of ill newborns, enriched for African American and rural patients. Our results demonstrate the utility of GS as it provides early in life detection of clinically relevant genetic variation not identified via current standard clinical testing, particularly for newborns exhibiting certain phenotypic features.

Many individuals with rare monogenic disease remain molecularly undiagnosed due to challenges accessing genetic testing, ambiguity in interpretation of uncertain variants, and latency between novel disease-gene discovery and adoption into clinical pipelines. The Rare Genomes Project (RGP) provides a remote, research-based genomic sequencing (GS) service for rare disease families to overcome diagnostic barriers while advancing the understanding of the genetic basis of rare disease. Short-read GS and family-centered analysis were performed to identify causes of primary disease. Variants meeting return criteria were clinically confirmed and returned to families. Since 2017, RGP enrolled clinically heterogeneous, undiagnosed rare disease families from all fifty states, Washington DC, and Puerto Rico. Findings were returned to 272 of 1014 sequenced families, resulting in a solve rate of 23.3% and a returned strong candidate rate of 3.5%. Condition-specific return rates ranged from 15% to 40% with many families receiving a result in a known disease gene after healthcare access challenges. Research interventions contributed to returns in 16% of all families and included variants in recently discovered or non-coding disease genes, additional functional studies to resolve variant uncertainty, and real-time data sharing to empower novel disease-gene discovery. RGP demonstrates that a research-based, patient-driven remote sequencing study can effectively bridge a diagnostic gap for rare disease families underserved in current clinical care. This study also underscores the clinical utility of GS with frequent re-analysis and highlights the critical continued role of research in variant interpretation, method innovation, and genomic knowledge base expansion for rare disease diagnosis.

Several medical societies including the American College of Medical Genetics and Genomics, the American Academy of Neurology, and the Association of Molecular Pathology recommend chromosomal microarray (CMA) as the first-tier test in the genetic work-up for individuals with neurodevelopmental disorders such as developmental delay and intellectual disability, autism spectrum disorder, as well as other disorders suspected to be of genetic etiology. Although CMA has significantly increased the diagnostic yield for these disorders, limitations in the technology preclude detection of certain structural variations in the genome and requires reflexing to other cytogenomic and molecular methods. Optical genome mapping (OGM) is a high-resolution technology that utilizes ultra-high molecular weight DNA, fluorescently labeled at a hexamer motif found throughout the genome, to create a barcode pattern, analogous to G-banded karyotyping, that can detect all classes of structural variations at very high resolution by comparison to a reference genome. A multisite study, partially published previously, with a total of n=1037 datapoints was conducted and showed 99.6% concordance between OGM and standard-of-care (SOC) testing for completed cases. The current phase of this study included cases from individuals with suspected genetic conditions referred for cytogenomic testing in a prospective postnatal cohort (79 cases with OGM and SOC results) and a retrospective postnatal cohort (262; same criteria). Among these cohorts were an autism spectrum disorder cohort (135) group with negative or uninformative results on previous testing (72). Prospective cases referred for CMA were included in this study as an unbiased comparison, OGM results show 100% concordance with variants of uncertain significance, pathogenic variants, and likely pathogenic variants reported by CMA other SOC and found reportable variants in an additional 10.1% of cases. Among the autism spectrum disorder cohort, OGM found reportable variants in an additional 14.8% of cases. Based on this demonstration of the analytic validity and clinical utility of OGM by this multi-site assessment, and considering clinical diagnostics often require iterative testing for detection and diagnosis in postnatal constitutional disorders, OGM should be considered as a first-tier test for neurodevelopmental disorders and/or suspicion of a genetic disease.

Patients with rare genetic disorders often wait years for a correct diagnosis, highlighting the urgent need for early and efficient triage for those at risk. Electronic health records (EHRs) are rich in useful information, yet the extent to which this data is sufficient to generate prediagnostic signals is uncertain. We trained machine learning models on billing codes and phenotypes extracted from clinical notes and applied these across eight conditions in a longitudinal cohort of roughly 3 million patient records from the Mayo Clinic, seeking early identification of conditions. The number of patients identified early varied by condition, ranging from 10% to 89% at 99% specificity. Median lead times exceeded one year prior to the first diagnostic billing code for most conditions, including Fabry disease (1,408 days), hereditary angioedema (2,755 days), neurofibromatosis type 1 (1,067 days), and hereditary hemorrhagic telangiectasia (2,320 days). We show that ML models trained on EHR data can prioritize patients for clinician review and confirmatory genetic testing and that the integration of diverse phenotypic data sources provides superior predictive value. Keywords: rare disease, electronic health records, phenotype risk score, machine learning, early diagnosis, Human Phenotype Ontology

BackgroundGenetic disease is common in the Level IV Neonatal Intensive Care Unit (NICU), but neonatology providers are not always able to identify the need for genetic evaluation. We trained a machine learning (ML) algorithm to predict the need for genetic testing within the first 18 months of life using health record phenotypes. MethodsFor a decade of NICU patients, we extracted Human Phenotype Ontology (HPO) terms from clinical text with Natural Language Processing tools. Considering multiple feature sets, classifier architectures, and hyperparameters, we selected a classifier and made predictions on a validation cohort of 2,241 Level IV NICU admits born 2020-2021. ResultsOur classifier had ROC AUC of 0.87 and PR AUC of 0.73 when making predictions during the first week in the Level IV NICU. We simulated testing policies under which subjects begin testing at the time of first ML prediction, estimating diagnostic odyssey length both with and without the additional benefit of pursuing rGS at this time. Just by using ML to accelerate initial genetic testing (without changing the tests ordered), the median time to first genetic test dropped from 10 days to 1 day, and the number of diagnostic odysseys resolved within 14 days of NICU admission increased by a factor of 1.8. By additionally requiring rGS at the time of positive ML prediction, the number of diagnostic odysseys resolved within 14 days was 3.8 times higher than the baseline. ConclusionsML predictions of genetic testing need, together with the application of the right rapid testing modality, can help providers accelerate genetics evaluation and bring about earlier and better outcomes for patients.

Background: The ClinGen ENIGMA BRCA1/BRCA2 Variant Curation Expert Panel (VCEP) has adapted the ACMG/AMP framework into gene-specific specifications. However, applying these specifications manually remains labour-intensive and prone to inconsistency, requiring integration of population, computational, functional, and clinical evidence through gene-specific decision trees and a points-based classification system. Methods: We developed HECTOR, a free web-based tool that implements the complete ENIGMA VCEP v1.2 specifications for BRCA1 and BRCA2. HECTOR automatically populates all evidence codes derivable from public data, routes curator-dependent evidence to a manual input layer and returns a transparent five-tier classification with code-level evidence. We validated HECTOR against two independent reference datasets: the 143-variant ENIGMA Evidence Repository, used as a clinical-grade reference standard, and 134 manually curated in-house variants of uncertain significance. HECTOR was then applied to the complete ClinVar BRCA1/BRCA2 catalogue (n = 34,077). Results: At the criterion level, HECTOR exactly reproduced 326 of 413 VCEP-assigned criteria (78.9%). The discordance arising predominantly from curator-dependent evidence rather than implementation errors whereas computationally accessible criteria showed perfect concordance. Across ClinVar, HECTOR classified 33,913 variants (99.5%). Agreement with definitive ClinVar classifications was 96.7% for pathogenic variants overall. Among variants for which HECTOR generated a definitive classification, directional concordance reached 99.7% for pathogenic and 99.9% for benign variants. HECTOR also resolved a substantial proportion of variants classified as uncertain (67.3%) or conflicting (88.7%), predominantly toward benign classifications. Conclusions: HECTOR provides a faithful, transparent implementation of the ENIGMA VCEP v1.2 specifications for BRCA1 and BRCA2, enabling rapid, standardized, and reproducible application of gene-specific variant classification guidelines while reducing the burden of manual curation.

Background: Rare genetic diseases collectively affect hundreds of millions of people, yet their clinical features are usually defined in small and selected patient groups which do not represent the entire population. The FMR1 CGG repeat expansion, one of the most common trinucleotide disorders, illustrates what we know about a genetic carrier's susceptibility that predisposes to multiple medical problems, which unfortunately remains poorly characterised. FMR1 premutation-associated conditions (FXPAC) encompass a spectrum of clinical manifestations. FMR1 premutation carriers are at risk of developing conditions such as fragile X-associated primary ovarian insufficiency (FXPOI), fragile X-associated tremor/ataxia syndrome (FXTAS), and a range of neuropsychiatric symptoms collectively termed Fragile X-associated neuropsychiatric disorders (FXAND), which to this date remain a topic of concern and debate. Despite an estimated carrier prevalence of 1 in 151 females and 1 in 468 males, FMR1 premutation carriers frequently remain undiagnosed or misdiagnosed. Electronic Health Records (EHRs) structured in the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) enable large-scale retrospective phenotyping of such rare conditions. Methodology: Using OMOP-mapped EHR data from the Mount Sinai Health System, we identified a cohort of 750 individuals with a confirmed FMR1 premutation status (98.5% female; median age at index 33.3 years), to our knowledge the largest EHR-derived FMR1 premutation cohort described within a single health system to date. Results: Several key FXPAC-associated diagnoses were documented before formal FMR1 premutation identification: Endocrine/Metabolic (16.5%, n = 124/750) and Neuropsychiatric (16.3%, n = 122/750) conditions were the most prevalent FXPAC domains. In every domain a substantial share of patients had their first record before the index date: the first-quartile lead time was 1.9 years for FXAND (n = 122), 1.5 years for FXPOI (n = 64), 1.0 years for both pain/fatigue (n = 19) and autoimmune/inflammatory conditions (n = 15), and 0.1 years for endocrine/metabolic conditions (n = 124); only FXTAS (n = 13) showed no pre-index lead (Q1 = 0.0 years). These lags reveal a measurable diagnostic gap prior to the identification of the FMR1 premutation. This study demonstrates that large-scale EHR-based phenotyping can characterise the clinical burden of rare genetic conditions, quantify diagnostic delays at the population scale, and identify precursor diagnosis patterns that may support earlier genetic referral in clinical practice.