Genomewide rapid association using mixed model and regression: a fast and simple method for genomewide pedigree-based...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Pedigree-based quantitative trait loci (QTL) association analysis commonly uses within-family variation (e.g., Transmission Disequilibrium Test, TDT).
• Measured genotype (MG) methods leverage both within- and between-family variation for increased power but are computationally intensive, limiting genomewide application.

Purpose of the Study:

• To introduce a computationally efficient method, genomewide rapid association using mixed model and regression (GRAMMAR), for genomewide pedigree-based QTL association analysis.
• To evaluate the performance of GRAMMAR compared to TDT-based and MG approaches in terms of statistical power, type 1 error, and speed.

Main Methods:

• GRAMMAR first computes residuals adjusted for family effects.
• It then employs rapid least-squares methods to analyze associations between these residuals and genetic polymorphisms.
• Selected polymorphisms can be further analyzed using the full MG approach.

Main Results:

• GRAMMAR demonstrates power comparable to the MG approach for moderately heritable traits (30%) in human pedigrees, significantly outperforming TDT-based methods.
• While MG may be slightly more powerful for very high heritabilities and large sibships, GRAMMAR shows minimal empirical power difference.
• GRAMMAR is substantially faster than MG, facilitating the analysis of hundreds of thousands of genetic markers.

Conclusions:

• GRAMMAR provides a computationally efficient and powerful alternative for genomewide pedigree-based QTL association studies.
• The method significantly enhances the feasibility of large-scale genetic analyses in complex pedigrees.
• GRAMMAR balances statistical power with computational speed, making it suitable for broad application in genetic research.