ProteoCast: a web server to predict, validate, and interpret missense variant effects.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• Predicting variant effects on protein function is crucial but challenging, especially for uncharacterized variants and intrinsically disordered regions.
• Existing computational tools often lack transparency and robust quality assessment.

Purpose of the Study:

• To develop ProteoCast, a web server for transparent and accurate prediction of variant effects on protein function.
• To identify functional linear peptides within intrinsically disordered protein regions.

Main Methods:

• ProteoCast utilizes evolutionary constraint analysis and structural context integration for variant effect prediction.
• It incorporates quality controls for multiple sequence alignments to ensure prediction reliability.
• A novel segmentation approach based on mutational sensitivity identifies functional motifs in disordered regions.

Main Results:

• ProteoCast achieves 77% sensitivity and 87% specificity in predicting pathogenicity for ClinVar variants.
• It demonstrates high accuracy across species, with 85% accuracy on Drosophila lethal mutations.
• The tool identifies twice as many functional motifs in intrinsically disordered regions compared to traditional phylogenetic methods.

Conclusions:

• ProteoCast offers a user-friendly, transparent, and efficient platform for variant effect prediction and functional site discovery.
• Its ability to analyze specific conformations enhances prediction accuracy and interpretability.
• The tool democratizes access to advanced variant analysis for the research community.