Genomic Databases

Genomic databases are essential tools for variant interpretation, providing curated information on variant pathogenicity, population frequency, gene-disease relationships, and functional evidence. Effective use of these databases is a core competency for clinical geneticists and genetic counselors interpreting molecular test results.

• ClinVar (NCBI): A freely accessible public archive of reported relationships between human genetic variants and phenotypes, with supporting evidence. Submissions come from clinical laboratories, research groups, and expert panels. Variants are assigned review status stars (0–4 stars) reflecting the level of curation: 0 stars = single submitter with no assertion criteria; 4 stars = reviewed by expert panel. Use for: checking whether a variant has been previously classified and by whom. Watch for conflicting interpretations between submitters.

• gnomAD (Genome Aggregation Database): Contains allele frequency data from >140,000 exomes and >76,000 genomes from individuals ascertained for not having severe pediatric disease. Use for: determining whether a variant is too common in the general population to be disease-causing. Key ACMG criteria: BA1 (benign stand-alone) if allele frequency >5% in any population; BS1 (strong benign) if frequency exceeds a disease-specific threshold; PM2 (supporting pathogenic) if absent or extremely rare. Important: gnomAD is not a "healthy" cohort; it may include individuals with late-onset conditions.

• OMIM (Online Mendelian Inheritance in Man): A comprehensive, curated catalog of human genes and genetic disorders. Each entry includes clinical features, molecular genetics, inheritance pattern, and key references. Use for: understanding gene-disease relationships, inheritance patterns, and clinical phenotypes. OMIM numbers: gene entries start with *, phenotype entries with #, gene/phenotype with +.

• HGMD (Human Gene Mutation Database): A curated collection of published germline mutations associated with human disease. Includes missense, nonsense, splicing, regulatory, and structural variants. Use for: identifying previously reported pathogenic variants and the associated literature. Note: the publicly available version is limited; full access requires a paid subscription. Not all entries are truly pathogenic; older entries may lack rigorous curation.

• ClinGen (Clinical Genome Resource): An NIH-funded initiative that defines the clinical relevance of genes and variants for precision medicine. Key outputs include: (1) Gene-disease validity classifications (definitive, strong, moderate, limited, disputed, refuted); (2) Dosage sensitivity maps (haploinsufficiency and triplosensitivity scores); (3) Variant curation expert panels (VCEPs) that provide definitive variant classifications in ClinVar; (4) Actionability assessments. Use for: determining whether a gene has sufficient evidence to be associated with a disease, and for expert-curated variant classifications.

• Population frequency thresholds (general guidance):

  • BA1 (benign stand-alone): >5% in any gnomAD population
  • BS1 (strong benign): Exceeds disease-specific maximum credible allele frequency (calculated based on disease prevalence, genetic heterogeneity, and penetrance)
  • PM2 (supporting pathogenic): Absent from or at extremely low frequency in gnomAD (below maximum credible allele frequency)

• Other useful resources:

  • DECIPHER: Database of genomic variants associated with developmental disorders; integrates phenotype and CNV data
  • UniProt: Protein sequence, structure, and function data
  • SpliceAI: Deep learning tool for predicting splice-altering variants
  • LOVD (Leiden Open Variation Database): Gene-specific databases curated by expert communities
  • Franklin (by Genoox): AI-assisted variant classification platform that aggregates data from multiple databases

• Variant interpretation workflow: A typical workflow queries ClinVar first (has this variant been seen before?), then gnomAD (how common is it?), then OMIM/ClinGen (is this gene associated with the patient's phenotype?), then HGMD and literature (what is the published evidence?).
• Resolving conflicting classifications: When ClinVar shows conflicting interpretations, examine the evidence provided by each submitter and prioritize expert panel reviews (3–4 star entries).
• Reclassification: Databases are updated continuously. A VUS classified 3 years ago may now have sufficient evidence for reclassification. Periodic reanalysis of negative or VUS-containing reports is recommended.

• ClinVar submissions vary in quality: Not all submissions undergo rigorous review. Older submissions may use outdated criteria. Always check the review status (star rating) and the evidence provided.
• gnomAD has ascertainment bias: Enriched for adult-onset diseases since pediatric severe disease cohorts were excluded. Some populations are underrepresented, affecting frequency estimates in diverse patients.
• HGMD includes false positives: Some variants listed as "disease-causing" in HGMD have been subsequently reclassified as benign. Always verify with current evidence.
• OMIM is not a diagnostic tool: It describes gene-disease associations and phenotypes but does not classify individual variants.
• No single database is sufficient: Accurate variant interpretation requires integrating evidence from multiple sources.