An efficient variance component approach implementing an average information REML suitable for combined LD and...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Variance component (VC) approaches using restricted maximum likelihood (REML) are established for quantitative trait loci (QTL) mapping.
• Incorporating linkage disequilibrium (LD) enhances mapping resolution but complicates standard Average Information (AI) REML algorithms due to dense covariance structures and potential numerical instability with high marker densities.

Purpose of the Study:

• To investigate a direct application of the variance-covariance matrix of all observations within AIREML for LD-based QTL mapping in complex pedigrees.
• To develop a more computationally efficient and numerically robust method compared to traditional mixed model equation-based approaches.

Main Methods:

• Utilized direct variance-covariance matrix analysis within AIREML for LD mapping.
• Applied the method to general complex pedigrees with high marker densities.
• Compared computational efficiency and numerical stability against mixed model equation-based methods.

Main Results:

• The proposed method demonstrates increased efficiency over standard mixed model equation approaches.
• The technique is robust to numerical issues arising from near-singularity caused by closely linked markers.
• Facilitates simultaneous fitting of multiple QTL, a significant computational advantage over existing methods.

Conclusions:

• Direct use of the variance-covariance matrix in AIREML offers a superior approach for LD-based QTL mapping.
• The method provides enhanced computational efficiency, numerical stability, and the capability for multi-QTL analysis in complex genetic architectures.