An integrative framework for clinical diagnosis and knowledge discovery from exome sequencing data

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Non-silent genetic variants, including nonsense mutations and insertion-deletion variants, significantly impact protein function and length but are often misclassified.
• Current variant effect prediction tools exhibit limited sensitivity and specificity for these critical variant types, hindering clinical applications.

Purpose of the Study:

• To develop and validate a novel method, Pathogenic Mutation Prediction (PMPred), for accurately predicting the pathogenicity of single nucleotide variations.
• To address the limitations of existing tools in classifying variants that cause premature protein termination and significant functional alterations.

Main Methods:

• Leveraging an ensemble machine learning model (UniGOPred) to monitor functional effects (Gene Ontology annotation changes) resulting from sequence variations.
• Identifying mutations with significant functional deviations from the wild-type sequence.
• Conducting comparative docking studies to validate pathogenicity predictions based on altered binding affinities.

Main Results:

• PMPred demonstrates increased sensitivity and specificity compared to state-of-the-art methods, particularly for variants causing substantial functional changes in proteins.
• Novel harmful mutations were identified in patient data, serving as motivating case studies.
• A comparative docking study confirmed PMPred's ability to correctly predict pathogenicity for a misclassified variant where other tools failed.

Conclusions:

• PMPred offers a more accurate approach to predicting the pathogenicity of single nucleotide variations, especially those with significant functional consequences.
• The method shows promise for improved clinical applications by reducing misclassification of critical genetic variants.
• PMPred is available as a free web service, facilitating broader research and clinical utility.