Unresolved rare disease represents a major global health challenge: approximately 300 million people are affected, and at least 50% of individuals remain genetically unresolved after standard diagnostic sequencing such as exome sequencing or whole genome sequencing. One recognized source of these missing genetic diagnoses is rare variation in the non-coding genome. Although whole genome sequencing can detect non-coding variants, these variants are frequently not interpreted in clinical pipelines, leaving potential causal variation unexplored.
To address this gap, the authors developed the Genomic Analysis of Variants in Unresolved Rare Disease (GAVURD) system. GAVURD is designed to systematically evaluate candidate pathogenic non-coding variants and produce a short, actionable list of prioritized variants for a given proband. The system integrates genomic variant calls, three-dimensional genome organization, and phenotype information to inform candidate selection.
GAVURD uses trio whole genome sequencing alignment data as its primary input. The system applies best-practice approaches for identifying both de novo variants and rare inherited variants from trio data. From the alignment files, GAVURD identifies candidate non-coding variants that are rare and that meet inheritance or de novo criteria relevant to rare disease genetics. The output is a focused set of non-coding variants for downstream linking and prioritization rather than an exhaustive list of all detected variants.
A core feature of GAVURD is the use of topologically associated domain (TAD) data to link non-coding variants to potential target genes. TADs represent regions of the genome with enriched internal chromatin interactions; variants within a TAD can potentially influence the expression of genes located elsewhere within the same domain. By harnessing TAD boundaries and domain assignments, GAVURD associates non-coding variants with human disease genes in a manner that accounts for three-dimensional genome architecture rather than relying solely on linear proximity.
This approach aims to improve the biological plausibility of variant–gene assignments for non-coding variants, helping to identify candidate regulatory variants that might alter gene expression relevant to the proband’s phenotype.
After variants are linked to candidate genes through TAD information, GAVURD rank-prioritizes variants based on phenotypic overlap. The system assesses concordance between the clinical features observed in the proband and known disease associations or phenotypes for the linked genes. Variants that both map to genes with relevant disease associations and match the proband phenotype receive higher priority.
This combined genotype–phenotype ranking is intended to produce a short, interpretable list of high-value candidates for experimental validation or further functional studies, facilitating efficient follow-up by diagnostic teams or research laboratories.
As a proof-of-concept demonstration, the authors applied GAVURD to ten probands with unresolved rare disease. From this application, GAVURD implicated six potentially causal non-coding variants based on a confluence of evidence supportive of pathogenicity. The report frames these implicated variants as candidates that merit additional functional follow-up to assess causality more directly.
The authors position this pilot application as evidence that GAVURD can generate plausible candidate non-coding causal variants from trio whole genome sequencing data in cases that remained unresolved after standard approaches.
GAVURD is presented as a tool to prioritize candidate non-coding causal variants, producing a manageable list that can guide targeted functional experiments. The authors emphasize the role of GAVURD in providing an informed starting point for experimental validation rather than delivering definitive clinical diagnoses on its own.
This work is reported as a preprint and has not been peer reviewed. Specific methodological details, performance metrics, and variant-level evidence beyond the high-level summary were reported in the preprint; readers interested in implementation or variant-level results should consult the full preprint and supplementary materials. The authors declared no competing interests and noted NIH training grant support in the funding statement.
Overall, GAVURD demonstrates a structured approach that integrates non-coding variant detection from trio whole genome sequencing, assignment to genes via TAD data, and phenotype-based ranking to prioritize candidates for unresolved rare disease cases. The system is intended to narrow the search space for functional follow-up studies and to augment diagnostic efforts in genetically unresolved patients.