Protein variant analysis, in one free report

What's new (September 30, 2026)

Genetic Disease Atlas

Search a genetic disease to see the human proteins linked to it and their variants, with ClinVar classifications and supporting evidence. Conflicting disease labels are flagged.

• Long QT syndrome genes and variants
• Hypertrophic cardiomyopathy genes and variants
• Marfan syndrome genes and variants
• Cystic fibrosis genes and variants
• Lynch syndrome genes and variants

AlphaGenome Atlas predictions

Now in every result. New jobs fill in 1-2 hours after they finish.

Clearer predictor results

Scores read against published thresholds, with a plain-language summary and rank.

Anonymous feedback

Every results page has a feedback form.

More variant formats

Frameshifts, UniProt-qualified variants and RefSeq transcript changes.

Evidence you can inspect

• AlphaMissense and REVEL computational predictions, with CADD and ESM1b signals when present
• ClinVar clinical evidence and gnomAD population frequencies
• 3D protein structure, structural neighborhoods, and hotspot analysis
• Experimental measurements, disease and phenotype associations, and published literature

Computational predictions support prioritization and review; they are not clinical diagnoses.

Public CATVariant analyses

Public pages summarize completed human protein analyses and link each gene to a stable page that can be reviewed or shared.

• 1,081 public human protein analyses
• 1,881,000+ analyzed variants
• 1,061 analyses with 3D structure
• 1,081 analyses with literature evidence
• 1,056 analyses with population data
• 1,081 analyses with experimental data

Examples from public gene pages

• KCNH2 — Long QT syndrome context; 2,334 public variants
• SCN9A — Epilepsy context; 4,096 public variants
• LMNA — Cardiomyopathy context; 1,553 public variants
• MYBPC3 — Cardiomyopathy context; 2,593 public variants
• KCNQ1 — Long QT syndrome context; 1,718 public variants

What a report shows

• Sequence: position, domains and nearby variants
• Structure: 3D location and variant hotspots
• Contacts: residues a variant touches, near and far
• Population: allele frequency across ancestry groups from gnomAD
• Predictors: 13 variant effect predictors, read against published thresholds
• Literature: papers from PubMed and LitVar that mention the variant

How it works

1. Enter a protein: gene symbol, UniProt ID or sequence.
2. Evidence is gathered from 12 public databases and 13 predictors.
3. Explore one report: sequence, 3D structure, tables and sources, linked.

Frequently asked questions

What is CATVariant?

A free web server for protein variant analysis. Enter a gene, UniProt ID or protein sequence, and CATVariant maps its variants onto the sequence and 3D structure together with clinical, population, experimental, predictor and literature evidence.

What can I enter?

A gene symbol, a UniProt accession or a protein sequence. Optionally add a list of variants (A561V, p.Ala561Val, frameshifts) or your own structure file.

Do I need a protein structure?

No. Experimental structures from the PDB or a predicted model from AlphaFold DB are found automatically.

Does it say whether a variant causes disease?

No. It brings clinical records, population data, assays and predictors together, each with its source. Clinical interpretation should follow ACMG/AMP guidelines and involve qualified professionals.

Where does the evidence come from?

UniProt, PDB, AlphaFold DB, gnomAD, ClinVar, ClinGen, dbSNP, MaveDB, LitVar, Europe PMC and Open Targets, plus predictors including AlphaMissense, REVEL, CADD, ESM1b and AlphaGenome.

What you can start with

Start with a gene symbol, UniProt accession, protein sequence, structure, or custom list of variants. CATVariant maps the submitted changes to protein sequence and structure, then organizes predictions, frequencies, clinical records, experimental measurements, and literature for follow-up.