GnomXMiss: AlphaMissense variants enriched with annotations from gnomAD, VEP, and ClinVar

File descriptions

GnomXMiss per-protein files

<uniprot_id>_GnomXMiss.tsv

Files contain missense variants for a UniProt canonical protein from the original Alphamissense data dump, enriched with:

gnomAD allele frequencies (overall + subpopulation-specific)

VEP annotations (GENCODE v46)

ClinVar variant metadata

These files are sorted by residue_number and include substitutions that were present in the AlphaMissense hg38 missense data dump and for which matching genomic coordinates could be retrieved in the gnomad.joint.v4.1.sites.ht table, enriched with external annotations where available.

Column descriptions

Below is the full set of columns that appear across the *_GnomXMiss.tsv files

🔹 Primary identifiers

uniprot_id

Canonical UniProtKB accession for the protein associated with the variant.

#CHROM

Chromosome name (e.g., chr1, chr2, chrX, chrY) based on GRCh38.

POS

1-based genomic coordinate of the variant (GRCh38).

REF, ALT

Reference and alternate nucleotides from gnomAD.

🔹 Protein-level variant information

protein_variant

Amino-acid substitution in the format <RefAA><Position><AltAA>, e.g. V600E.Derived from AlphaMissense or imputed using VEP when missing.

residue_number

Integer position of the substitution extracted from protein_variant.

🔹 AlphaMissense annotation

am_class

Classification of variant pathogenicity:

likely_benign

ambiguous

likely_pathogenic

am_pathogenicity

Calibrated AlphaMissense pathogenicity score (0–1).

(Exact integration pipeline described in Zenodo README.)

🔹 ClinVar annotations (if available)

existing_variation

ClinVar or dbSNP IDs extracted via Ensembl VEP.

clinvar_significance

e.g., Benign, Likely_pathogenic, VUS, etc.

clinvar_condition

Name(s) of associated disease conditions.

clinvar_disease_id

Disease database identifiers (MEDGEN, MONDO, OMIM, etc.).

clinvar_variant_id

ClinVar Variation ID.

🔹 gnomAD allele frequency data (v4)

These fields are derived from the joint gnomAD v4 release and represent population-level statistics (no individual data).

Overall

allele_count

allele_number

allele_frequency

homozygote_count

Population-specific allele frequencies

The dataset includes AF values for multiple ancestry groups:

nfe_allele_frequency — Non-Finnish European

fin_allele_frequency — Finnish

afr_allele_frequency — African/African American

eas_allele_frequency — East Asian

sas_allele_frequency — South Asian

amr_allele_frequency — Latino/Admixed American

ami_allele_frequency — Amish

asj_allele_frequency — Ashkenazi Jewish

mid_allele_frequency — Middle Eastern

remaining_allele_frequency — Remaining ancestry group

Sex-specific allele frequencies

XX_allele_frequency

XY_allele_frequency

📜 Copyright & Licensing

1. Copyright (this dataset)

All processed data files in this deposit (*_GnomXMiss.tsv) and all accompanying scripts are released under:

Creative Commons Attribution 4.0 International (CC-BY-4.0)© 2025 Chataigner Lucas

You are free to:

Share and Adapt 

Under the terms:

Attribution must be given to the original authors and sources (see below).

This work is provided “AS IS” with no warranty.

2. Attribution for Upstream Sources

This dataset is a derived work built from several openly licensed public resources.No raw datasets are redistributed, only derived, merged annotations.

AlphaMissense (DeepMind)

Source predictions © DeepMind Technologies LimitedLicensed under Creative Commons Attribution 4.0 International (CC-BY-4.0).You must cite:

Cheng et al. Accurate proteome-wide missense variant effect prediction with AlphaMissense.CC-BY-4.0.https://doi.org/10.1038/s41586-023-06668-0

AlphaMissense licensing and disclaimer apply:“Not intended for clinical use.”

gnomAD (Broad Institute / gnomAD Consortium)

This deposit includes derived allele frequency statistics only extracted from the gnomAD v4 joint release.gnomAD is licensed under CC-BY-4.0.Citation:

Karczewski et al. The mutational constraint spectrum quantified using variation in 141,456 humans.Nature (2020). doi:10.1038/s41586-020-2308-7

No raw gnomAD data are redistributed.

ClinVar (NCBI / NIH / NLM)

ClinVar data are in the public domain (CC0).This deposit redistributes only derived annotations (significance, condition, disease IDs, variant IDs).

Citation:

Landrum et al. ClinVar: improving access to variant interpretations and supporting evidence.Nucleic Acids Research (2018). doi:10.1093/nar/gkx1153

NIH disclaimer applies:Not intended for clinical decision making.

Ensembl VEP & GENCODE

This dataset includes only VEP-derived annotations, not the software itself.VEP is © EMBL-EBI, distributed under GPLv3.GENCODE v46 transcript models are used for annotations.

Citations:

McLaren et al. The Ensembl Variant Effect Predictor. Genome Biology (2016).Frankish et al. GENCODE reference annotation for the human genome.

UniProt

UniProt identifiers are free to use and redistribute.

Citation:

UniProt Consortium. UniProt: the Universal Protein Knowledgebase. Nucleic Acids Research (2023).