henhaobio
Sign in

Nucleic Acids ResearchBioinformaticsPeer-reviewed journal article

MetaDome 2027: a comprehensively updated resource for aggregating missense variant evidence across homologous human protein domains

Laurens WieliD, Federico FerraroiD, Jay Yu, Jimmy ZheniD, Daniel C. NachuniD, Rodrigo MendeziD, Chloe M. ReuteriD, Jason L. CuiiD, Devon E. BonneriD, Jennefer N. CarteriD, Shruti MarwahaiD, Maartje van de VorstiD, Sara EmamiiD, Elijah KravetsiD, Matthew B. NeuiD, Tjakko J. van HamiD, Tjitske KleefstraiD, Euan Angus AshleyiD, Jonathan A. BernsteiniD, Stephen B. MontgomeryiD, Christian GilisseniD, Matthew Thomas WheeleriD

PDFDOIPreprintCodeData

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (31 Aug 2026) · CC BY 4.0

MetaDome now supports GRCh38 alongside GRCh37, adds coordinate lookup, and identifies 52,463 GRCh38 ClinVar missense variants of uncertain significance with uncontradicted pathogenic evidence at homologous Pfam-domain positions.

Most missense variants are observed too rarely to interpret individually; pooling gnomAD and ClinVar variation across equivalent positions of homologous Pfam domains (meta-domains) amplifies the per-position signal and is now named in diagnostic classification guidelines. The 2019 MetaDome predated gnomAD's growth from ∼123,000 exomes to over 800,000 exomes and genomes and the shift to GRCh38; this update rebuilds the mappings on current releases for both assemblies and adds coordinate-based lookup.

Key results

  1. ShownAcross ClinVar 2025-10-06 on GRCh38, 147,835 in-domain missense variants of uncertain significance have pathogenic or likely pathogenic missense evidence at the equivalent meta-domain position; 52,463 of them have such evidence only at homologous positions, uncontradicted by benign variation.

    Of 1,920,156 missense variants of uncertain significance, 786,926 (41.0%) are in Pfam domains; among these, 147,835 (18.8%) show one or more pathogenic/likely pathogenic homologue variants. 3,337 (2.3%) are also pathogenic at the same residue; 52,463 (35.5%) homologous-only, uncontradicted; 92,035 (62.3%) conflicting. 25,407 rest on three or more Pathogenic variants. GRCh37: 140,374 of 765,455 (18.3%). · Illustrative Use-Case; Figure legends

  2. ShownThe GRCh38 build maps more genes, transcripts and Swiss-Prot sequences than MetaDome 2019, and Pfam families eligible for meta-domains (two or more human instances) rose 73.3% on the comparable GRCh37 build.

    GRCh38 build: 19,209 protein-coding genes (18,736 in 2019), 45,778 transcripts (42,116), 33,629 Swiss-Prot sequences (33,492) at 100% identity. GRCh37 build: Pfam instances 71,419 to 82,069 (+14.9%), eligible families 3,334 to 5,778 (+73.3%). Pathogenic/likely pathogenic ClinVar missense variants inside a meta-domain: 67.9% (37,692 of 55,548) on GRCh38; 65.4% on GRCh37 vs 72% reported in 2019. · Database growth; Tables

  3. SuggestedA diagnostic laboratory reclassified a de novo RALA p.Lys16Glu variant from uncertain significance to likely pathogenic. The support included pathogenic variants at the equivalent Ras-domain residue in the paralogues RIT1 and KRAS.

    Lys16 sits at a Ras Family (PF00071) consensus column also occupied by 124 other human codons; ClinVar lists six pathogenic/likely pathogenic missense variants there (RIT1: two Lys>Asn, one Lys>Glu, the identical change; KRAS: two Lys>Asn, one Lys>Gln). Applied as ACMG/AMP PM5 at moderate weight (LR+ = 7, Gunning and Wright 2023) with structural analysis and absence from gnomAD v4.1. · Illustrative Use-Case; MetaDome 2027 Web Platform Interface and…; Figure legends

  4. ShownThe rebuilt pipeline reproduces the published GRCh37 tolerance landscape almost exactly. Ambiguous transcript-to-Swiss-Prot assignments are rare, and every input row is either loaded or logged as skipped.

    99.88% of codons shared with the 2019 release keep an identical score. Multiple full-length-identical Swiss-Prot candidates arose for one of 45,778 transcripts (GRCh38) and seventeen of 43,143 (GRCh37), all identical-sequence paralogues. Gene-name mismatches with UniProt were flagged for 563 (1.23%) and 3,161 (7.33%) transcripts. Integrity checks returned zero violations. · Mapping pipeline; Tolerance landscape

  5. ShownCiting papers mostly use MetaDome for variant-level gene-disease association and clinical variant interpretation. The Association for Clinical Genomic Science (ACGS) 2024 germline guidelines and the SVIG-UK 2026 somatic guidelines name it as an evidence source.

    Of 317 citations (Google Scholar, April 2026), 261 were independent peer-reviewed uses. Primary category: gene-disease association 127 (49%), clinical interpretation 90 (34%); with all applicable categories counted, clinical interpretation appears in 201 (77%). 60 (23%) tied the evidence to an explicit ACMG/AMP criterion, most often PM1 (23). ACGS cites it under PM1; SVIG-UK under oncogenicity code O8. · Community Use of MetaDome; Adoption in clinical variant-interpretation…; Figure legends

Methods, data and reproducibility

A Python pipeline, run separately for GRCh37 and GRCh38, re-translates GENCODE coding sequences, pairs them with Swiss-Prot only at full-length identity (BLAST+), assigns Pfam domains (pfam_scan/HMMER), and precomputes HMM consensus-column links so gnomAD and ClinVar variation aggregates per meta-domain position; tolerance is a codon-corrected missense/synonymous ratio in 21-residue windows.

  • GENCODE protein-coding transcripts (v45 for GRCh38, v19 for GRCh37) · human · n = 45,778 transcripts (GRCh38); 43,143 (GRCh37)
  • UniProtKB/Swiss-Prot canonical and isoform sequences, release 2025_01 · human · n = 33,629 sequences mapped (GRCh38 build)
  • Pfam v37.4 domain instances in the human proteome · human · n = 87,814 instances across 5,908 families (both builds combined)
  • gnomAD population variation (v4.1 for GRCh38; r2.0.2 for GRCh37) · human · n = over 800,000 exomes and genomes (current gnomAD, per Introduction)
  • ClinVar release 2025-10-06 missense variants (pathogenic/likely pathogenic and uncertain significance) · human · n = 55,548 pathogenic/likely pathogenic; 1,920,156 uncertain (GRCh38)
  • Trio genome sequencing of a 12-year-old patient with developmental delay (RALA use case) · human · n = one patient (trio) · ClinVar VCV 2582518
  • Papers citing MetaDome 2019, reviewed in full text and assigned usage categories · n = 317 citations; 261 independent peer-reviewed uses
Methods & tools
BLAST+ v2.16.0, pfam_scan v1.6, HMMER 3.4, Pixi, Flask, SQLAlchemy 2.0, PostgreSQL v18, Celery v5, Docker Compose, D3.js v7.9.0, bcftools, bedtools
Data & accessions
http://www.metadome.app/
Software & environment
Build pipeline runs in a Pixi environment pinning every tool and dependency; platform uses Python 3.13, Flask, SQLAlchemy 2.0, PostgreSQL v18, Celery v5, Redis v8 and RabbitMQ v4 under Docker Compose; full version matrix in Supplementary Table S6. Code under MIT License.
Experimental conditions
Build hardware and runtime not stated. Front-end timing: the TTN dashboard loads in 10.2 s vs 48.6 s for the 2019 stack (median of four loads, MacBook Pro M1 Max, Chrome).
Reproducibility gaps
The Zenodo data release (mapping tables, per-residue dataset, genome-browser tracks, prebuilt output; CC-BY 4.0) is described, but the text prints no DOI or URL for it. The candidate sets are in Supplementary Data S2 and S3.
Limitations (5)
  • The counts of paralogue evidence are a lower bound. ClinVar records with the combined Pathogenic/Likely_pathogenic label are excluded, and identical amino-acid substitutions (PS1-type evidence) are not separated from PM5-type evidence. · stated by the authors
  • The proteome-wide analysis reclassifies no variant and does not report how many distinct genes contribute to each candidate's homologous evidence. It only locates evidence for laboratories to weigh alongside their other criteria. · stated by the authors
  • Meta-domains rest on purely sequence-based Pfam domain prediction. The authors expect the move to InterPro and AlphaFold-informed domain boundaries to refine them in future releases. · stated by the authors
  • The paper does not check the 52,463 uncontradicted candidates against any outcome. The single RALA reclassification also used structural analysis and absence from population data, so the accuracy that meta-domain evidence adds is not measured here. · noted in this summary
  • GRCh37 stays on gnomAD r2.0.2, so tolerance scores for the same residue differ between assemblies at almost all shared positions (2.17% identical). Results therefore depend on which assembly is chosen. · noted in this summary

Read this if you classify missense variants or build variant-effect tools and want per-residue gnomAD and ClinVar evidence pooled across paralogues, which you can query by genomic coordinate on GRCh38 or GRCh37 or download for reuse.

Citations
0
FWCI
0.00
Field percentile
top 41%

Abstract

The interpretation of missense variants remains a major challenge in clinical genetics. "Meta-domains" aggregate population and pathogenic variation across homologous Pfam domain instances in the human proteome, providing per-residue context for interpreting variants of uncertain significance (VUS). Our 2019 implementation, MetaDome, is widely used and named in clinical variant-classification guidelines. Here we present the MetaDome 2027 update, featuring a comprehensively updated dataset and GRCh38 support. The redesigned pipeline enables incremental updates of GENCODE, UniProtKB/Swiss-Prot, Pfam, gnomAD, and ClinVar while maintaining 100% sequence-identity gene-to-protein mapping. Annotated Pfam domain instances grew 14.9% from 71,419 to 82,069 and meta-domain-eligible Pfam families ([≥]2 human occurrences) by 73.3% from 3,334 to 5,778; Pfam domains are annotated to 92% of human proteins. Approximately 43% of mapped protein-coding nucleotides (14.3 million in GRCh38, 13.8 million in GRCh37) are in a meta-domain; in GRCh38 67.9% (37,692 of 55,548) of pathogenic or likely pathogenic ClinVar missense variants fall at such a position. We show how MetaDome helped reclassify a de novo missense VUS in RALA and identify 52,463 ClinVar missense VUS for which meta-domains supply otherwise unavailable pathogenic evidence. MetaDome is freely available at www.metadome.app.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISWiel '19MacGowan '17Beltran '24Kwon '26Wiel '22Pérez‐Palma '20
76% semantic matchBioinformatics
Laurens Wiel, Coos A. B. Baakman, Daan Gilissen+3 more·Human Mutation·May 22, 2019· 264 citations