Assessing variant effect predictors and disease mechanisms in intrinsically disordered proteins

Area of Science:

• Genomics
• Proteomics
• Computational Biology

Background:

• Intrinsically disordered regions (IDRs) lack stable structures, impacting cellular processes and disease variant interpretation.
• IDRs' dynamic nature and involvement in regulation pose challenges for predicting genetic variant effects.

Purpose of the Study:

• To systematically assess the distribution of pathogenic and benign missense variants across structured and disordered protein regions.
• To evaluate the performance of variant effect predictors (VEPs) on variants within IDRs.

Main Methods:

• Analysis of missense variant distributions across the human proteome, categorizing regions as disordered, intermediate, or structured.
• Systematic evaluation of 33 VEPs using metrics like sensitivity and AUROC scores for variants in different protein regions.

Main Results:

• Pathogenic variants are depleted in IDRs but linked to dominant gain- and loss-of-function mechanisms.
• VEPs show reduced sensitivity for pathogenic variants in IDRs, despite high overall accuracy driven by benign variant predictions.
• Significant discordance exists among VEP classifications for variants in disordered regions.

Conclusions:

• Current VEPs require refinement for accurate variant effect prediction in IDRs.
• Developing disorder-informed prediction strategies and region-aware thresholds is crucial for interpreting genetic variants in disease.
• Incorporating IDR-specific biological features may improve future variant prediction tools.