Missense3D-TM: Predicting the Effect of Missense Variants in Helical Transmembrane Protein Regions Using 3D Protein...

Area of Science:

• Genomics
• Structural Biology
• Bioinformatics

Background:

• Variant effect predictors typically focus on soluble proteins.
• Existing tools often fail to account for the unique environment of membrane proteins.
• Accurate prediction of variant effects in membrane proteins is crucial for understanding disease.

Purpose of the Study:

• To develop and validate Missense3D-TM, a structure-based tool for predicting missense variant effects in membrane proteins.
• To assess the performance of Missense3D-TM compared to existing methods.

Main Methods:

• Developed Missense3D-TM, a structure-based variant effect predictor specifically for membrane proteins.
• Evaluated Missense3D-TM on a dataset of pathogenic and benign variants from 706 membrane protein structures.
• Compared Missense3D-TM performance against mCSM-membrane.

Main Results:

• Missense3D-TM achieved 66% accuracy, 58% sensitivity, and 81% specificity on a large dataset.
• Performance was comparable to mCSM-membrane.
• The tool provides structural context and predicted membrane location for variants.

Conclusions:

• Missense3D-TM offers a valuable structure-based approach for assessing missense variants in membrane proteins.
• The tool's predictions can aid in understanding the pathogenicity of variants within the membrane environment.
• A web server is available for public use.