New Database Links Membrane Protein Variants

The research-stage preprint maps disease variants to membrane environments, offering a route toward mechanism-driven study.

Updated on Nov. 5, 2026 in Life Sciences

Flat-color editorial illustration of a membrane cross-section as horizontal bands, with a protein shape spanning them and small red dots marking variant sites.
Researchers released memVar, a preprint database linking about 10.9 million single-nucleotide variants across 7,732 human membrane proteins to their membrane environments. AI Illustration. Upload story photo >

memVar, a new database, has released an integrated set of 10,906,688 single-nucleotide variants across 7,732 human membrane-associated proteins. The work is a bioRxiv preprint, so it is research-stage and has not been peer reviewed.

Why it matters

Membrane proteins sit inside the lipid bilayer and touch many other molecules, so their disease variants are hard to interpret without knowing that environment. Connecting clinical, experimental and predictive data to membrane topology and interaction sites points toward mechanism-driven research on which variants actually break function.

The enrichment figures come from pooled odds ratios comparing pathogenic and benign variants across the database. The 57.9% versus 7.4% split shows that pathogenic substitutions were far more often recorded as disrupting a protein interaction, though this reflects only experimentally tested cases.

The players

memVar

An integrated membrane protein variation database linking variants from ClinVar, COSMIC and gnomAD to membrane topology and interaction data.

ClinVar

A public archive of clinical assertions about the relationship between human genetic variants and disease, maintained by the U.S. National Center for Biotechnology Information.

gnomAD

The Genome Aggregation Database, a large public catalogue of human genetic variation drawn from population sequencing data.

COSMIC

The Catalogue of Somatic Mutations in Cancer, a curated resource of somatic mutations observed in human cancers.

The details

memVar reconciles genomic, transcript and protein coordinates so that each variant can be placed on the protein it alters. It then links that position to membrane topology, bilayer depth, lipid contacts, functional sites and partner-specific interaction interfaces, along with variant-effect predictions. Coordinated sequence, structure and variant views keep track of where each piece of information came from and separate clinical assertions, experimental observations and computational predictions.

Timeline

  1. November 5, 2026: memVar preprint published on bioRxiv.

The Tech Race

Variant databases such as ClinVar, COSMIC and gnomAD catalogue which mutations occur and what they are linked to, but they do not record where a mutation sits within a membrane or which partners it touches. memVar adds those layers, placing it alongside existing resources as an attempt to make variant interpretation more mechanistic.

memVar is a research-stage preprint and is not a clinical tool, so it does not change any diagnosis or treatment today. It may help researchers prioritize which membrane protein variants to test in the lab, but that work is still ahead.

The takeaway

The key trajectory is that variant interpretation for membrane proteins may move from listing mutations toward explaining their molecular context. Watch for the peer-reviewed version of the preprint and whether experimental follow-up on the W202R variant of uncertain significance confirms effects on protein maturation, surface abundance or channel activity.

Further reading

For more on how genetic variation is studied across biology, explore Life Sciences.

More information

Explore the memVar membrane protein variation database directly.

Source note: This article includes information reported by Biorxiv.