Preparation and Curation of Phenotypic Datasets.

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Genome-wide association studies (GWAS) are crucial for identifying genetic variants associated with traits.
• The accuracy of quantitative trait loci (QTL) detection in GWAS is sensitive to the quality and analysis of phenotypic data.
• Phenotypic data curation and statistical modeling choices can influence the reliability of GWAS results.

Purpose of the Study:

• To investigate how phenotypic data curation and statistical analysis methods affect the identification of QTL in GWAS.
• To demonstrate the impact of outlier cleaning and different statistical methods for estimating genotypic means on GWAS outcomes.
• To provide recommendations for optimizing GWAS pipelines for more accurate QTL detection.

Main Methods:

• Case study analysis of existing datasets.
• Evaluation of the effect of outlier removal on QTL identification.
• Comparison of different statistical methods for estimating genotypic mean values of phenotypic data.
• Assessment of mixed models incorporating spatial trends for GWAS.

Main Results:

• Failure to clean outliers led to a higher number of dubious QTL, particularly at loci with unbalanced allelic frequencies.
• A trade-off exists between minimizing false positives and avoiding the loss of rare but significant alleles.
• Different statistical methods for estimating genotypic means resulted in reduced overlap in identified QTL.
• Mixed models enhanced trait heritability, increased QTL numbers, and improved overall GWAS power.

Conclusions:

• Phenotypic data cleaning, including outlier removal, is essential for reliable QTL detection in GWAS.
• Employing robust statistical models, such as mixed models, for estimating genotypic means improves GWAS accuracy.
• Integrating data cleaning and appropriate statistical modeling into GWAS pipelines minimizes both false positive and false negative rates for QTL detection.