The Complementarity Between Protein-Specific and General Pathogenicity Predictors for Amino Acid Substitutions

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing necessitates tools for assessing sequence variant functional impact.
• General methods (GMs) for single amino acid variants have limitations in accuracy (∼80%) for clinical use.
• Protein-specific predictors (PSPs) show potential to outperform GMs but face challenges like performance thresholds and data scarcity.

Purpose of the Study:

• To characterize the relationship between protein-specific predictors (PSPs) and general methods (GMs).
• To compare the performance of 82 derived PSPs against established GMs (PolyPhen-2, SIFT, PON-P2, MutationTaster2, CADD).
• To explore how the complementarity of PSPs and GMs can enhance pathogenicity prediction.

Main Methods:

• Derived 82 protein-specific predictors (PSPs).
• Compared PSPs against multiple general methods (GMs) including PolyPhen-2, SIFT, PON-P2, MutationTaster2, and CADD.
• Analyzed the performance relationship and complementarity between PSPs and GMs.

Main Results:

• A complementary relationship was observed between PSPs and GMs, with no single method consistently superior.
• The performance comparison varied, with some PSPs outperformed by GMs like PON-P2, while others outperformed GMs like SIFT.
• The study identified specific scenarios where each approach demonstrated relative strengths.

Conclusions:

• PSPs and GMs offer complementary information for variant effect prediction.
• The choice between PSPs and GMs depends on the specific protein and variant context.
• Combining PSPs and GMs holds promise for increasing the success rates in predicting variant pathogenicity.