About

Mechanistic interpretation of missense variants

Most missense variant predictors answer one question. Is this variant likely to be damaging. They do not explain how.

This resource links predicted pathogenicity to predicted mechanism. It combines sequence-based scores with structure-based annotations for protein stability, protein-protein interfaces, and small-molecule binding pockets.

The result is a structured view of variant impact that can be inspected, compared, and reused.

Where this fits

Variant discovery pipelines produce candidate variants. This resource operates after that step.

Variant → pathogenicity prediction → mechanistic annotation → interpretation

It does not change how variants are detected. It adds a mechanistic layer to support interpretation.

What this provides

  • pathogenicity scores and labels
  • sequence-model evidence
  • predicted stability change
  • interface annotation
  • pocket annotation
  • a mechanism label where supported

These features are presented together so the likely mode of disruption can be assessed directly.

What this does not do

This resource does not call variants.

It does not prioritise variants.

It does not provide clinical classification.

It does not replace experimental validation or expert review.

It provides mechanistic annotations as a data layer.

Source

Jänes et al. Predicted mechanistic impacts of human protein missense variants. bioRxiv (2024).

https://doi.org/10.1101/2024.05.29.596373

Related data is available via ProtVar at EMBL-EBI.

https://www.ebi.ac.uk/ProtVar/