Regulatory Single-Nucleotide Variant Predictor Increases Predictive Performance of Functional Regulatory Variants

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Identifying genetic variants associated with human inherited diseases is crucial.
• Existing bioinformatics tools often struggle with variants outside protein-coding regions.
• Regulatory variants play a significant role in disease but are less studied.

Purpose of the Study:

• To improve the classification performance of the regulatory single-nucleotide variant predictor (RSVP) for variants causing regulatory abnormalities.
• To enhance the prediction of pathogenic significance for rare or de novo variants.
• To develop a framework for assessing all functional regulatory variants.

Main Methods:

• Incorporated genomic regions identified by the ENCODE project into the RSVP tool.
• Compared RSVP performance with Genome-Wide Annotation of Variants (GWAVA) and Combined Annotation-Dependent Depletion (CADD).
• Utilized features related to the nearest gene, including expression and Gene Ontology (GO) terms.

Main Results:

• RSVP's performance for regulatory variants increased from an Area Under the Curve (AUC) of 0.90 to 0.97.
• RSVP demonstrated comparable performance to GWAVA but outperformed it on variants near the transcription start site (AUC: 0.96 vs. 0.71).
• RSVP's superior performance is attributed to the inclusion of nearest gene features.

Conclusions:

• The enhanced RSVP tool offers improved accuracy in predicting the functional impact of regulatory variants.
• The findings suggest a promising framework for assessing the pathogenicity of regulatory variants.
• This advancement aids in identifying genetic markers for human inherited diseases.