0%

​

Share this page with:

Annotations tab

This tab displays annotations from AlphaMissense, an AI model that predicts the pathogenicity of missense mutations in human proteins.

For canonical human proteins, you can now explore the predicted structures and toggle between two key visual representations: the quality of the structure model (pLDDT) and the average pathogenicity scores from AlphaMissense.

Annotations tab. The heatmap displays the predicted pathogenicity of missense variants, which can be filtered by category (likely pathogenic, likely benign, uncertain). Hovering over a specific variant in the heatmap shows its score and highlights the residue’s position on the 3D structure.

The variant data is displayed using an interactive heatmap and 3D visualisation, offering insights into the broader implications of specific residue changes. The heatmap can help with interpreting important functional areas with higher predicted pathogenicity. AlphaMissense categorises mutations as “Likely pathogenic”, “Likely benign”, or “Uncertain” and provides a score. You can filter the heatmap by category and modify the ranges.

You can also download the heatmap data and pathogenicity scores in CSV format from the main Download option at the top of the entry.

If there is no AlphaMissense data available for a particular protein, the heatmap won’t be displayed, but the Annotations tab will remain enabled. This allows you to still access other available annotations for that entry.

If you are interested in a deeper dive into AlphaMissense, you can find a dedicated training here: “AlphaFold practical guide: Classifying the effects of missense variants using AlphaMissense”. 

If you use data from AlphaMissense in your work, please cite the following paper: