Identification of Influential Variants in Significant Aggregate Rare Variant Tests

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Computational Biology

Background:

• Rare genetic variants play a role in complex diseases.
• Identifying specific causal variants within aggregate sets is challenging.
• Existing methods for variant prioritization have limitations.

Purpose of the Study:

• To introduce a novel method for prioritizing rare variants after a significant aggregate association test.
• To quantify the influence of individual variants on aggregate association tests.
• To identify a subset of variants that influence disease association.

Main Methods:

• Developed the Rare Variant Influential Filtering Tool (RIFT) using outlier detection methods (Tukey fences, median absolute deviation, SD).
• Evaluated RIFT's performance through simulations, comparing true and false positive rates against adaptive combination of p values (ADA) and BeviMed.
• Applied RIFT to targeted resequencing data from idiopathic pulmonary fibrosis (IPF) studies.

Main Results:

• RIFT and other outlier detection methods showed higher sensitivity for uncommon variants (0.001 < MAF < 0.03) than very rare variants (MAF < 0.001).
• RIFT demonstrated a lower median false positive rate than ADA for uncommon variants.
• Both ADA and RIFT achieved significantly higher true positive rates compared to BeviMed.
• In IPF data, 6 polymorphisms were identified as highly influential in disease association.

Conclusions:

• RIFT is an effective tool for prioritizing rare variants in association studies, offering a high true positive rate and low false positive rate.
• This method enhances the resolution of rare variant signals within aggregate sets.
• RIFT provides an objective basis for prioritizing variants for experimental validation and understanding biological pathways.