Population-aware permutation-based significance thresholds for genome-wide association studies

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Permutation-based significance thresholds offer a robust alternative to Bonferroni correction in genome-wide association studies (GWAS), especially with skewed phenotype distributions.
• The permGWAS method introduced efficient batch-wise computation but suffered from redundant calculations and disrupted population structure.
• Traditional permutation methods that only permute phenotypes can break underlying population structures, leading to inaccurate results.

Purpose of the Study:

• To develop an improved permutation-based GWAS method (permGWAS2) that preserves population structure and optimizes computational efficiency.
• To reduce redundant computations and enhance the accuracy of significance thresholds in GWAS.
• To identify novel genetic associations with adaptive traits and filter out false positives.

Main Methods:

• Implemented permGWAS2, a novel method utilizing block matrix decomposition for optimized permutation computations.
• Preserved population structure throughout the permutation process, unlike traditional methods.
• Validated permGWAS2 on synthetic datasets and re-analyzed a wild sunflower (Helianthus annuus L.) dataset with 86 phenotypic traits.

Main Results:

• permGWAS2 demonstrated a lower false discovery rate for skewed phenotypes compared to permGWAS and Bonferroni correction on synthetic data.
• Analysis of the wild sunflower dataset identified numerous novel associations with putatively adaptive traits.
• Several likely false-positive associations previously reported were successfully removed, increasing confidence in the findings.

Conclusions:

• permGWAS2 offers a significant advancement in GWAS methodology by maintaining population structure and improving computational efficiency.
• The method provides more accurate significance thresholds, leading to more reliable identification of genetic associations.
• permGWAS2 is a valuable open-source tool for genetic research, particularly for studies with complex population structures and skewed phenotypes.