Variant characterization in the intrinsically disordered human proteome

Area of Science:

• Genomics
• Computational Biology
• Structural Biology

Background:

• Variant effect prediction is crucial for precision medicine.
• Computational models excel at predicting effects in folded proteins but struggle with intrinsically disordered regions (IDRs).
• A significant portion of missense variants occur in IDRs, posing a challenge for accurate prediction.

Purpose of the Study:

• To develop and validate a method for predicting variant effects in intrinsically disordered protein regions (IDRs).
• To structurally annotate protein-protein interactions involving disordered motifs.
• To enable structure-based prediction of variant deleteriousness in IDRs.

Main Methods:

• Combined sequence pattern searches with AlphaFold to model protein-protein interaction interfaces.
• Selected interfaces overlapping with uncertain missense variants.
• Predicted variant effects in IDRs using structural models.
• Validated predictions through experimental methods.

Main Results:

• Structurally annotated 1,300 protein-protein interactions involving disordered motifs.
• Enabled prediction of deleterious effects for 1,187 variants in IDRs.
• Experimental validation confirmed predicted interfaces and variant effects.
• Identified deleterious variants missed by AlphaMissense, highlighting method's utility.

Conclusions:

• Integrating sequence analysis with structural modeling (AlphaFold) effectively predicts variant effects in IDRs.
• This approach generates testable hypotheses for variant function in intrinsically disordered proteins.
• Advances the characterization of variants in IDRs, crucial for precision medicine applications.