Comparison of mixed-model approaches for association mapping

Area of Science:

• Agricultural Science
• Genetics
• Biotechnology

Background:

• Traditional linkage-mapping methods have limitations in genetic studies.
• Association mapping offers a powerful alternative for dissecting complex traits.
• Understanding wheat (Triticum aestivum L.) genetics is crucial for crop improvement.

Purpose of the Study:

• To evaluate various association mapping methods in autogamous wheat using empirical data.
• To determine a robust marker-based kinship matrix using restricted maximum-likelihood (REML).
• To compare one-step and two-step association mapping approaches for trait discovery.

Main Methods:

• Phenotypic and genotypic data from 303 soft winter wheat inbreds were analyzed.
• A marker-based kinship matrix was estimated using REML.
• Spearman's rank correlation was used to compare P-values from different methods.
• Mixed-model association mapping approaches were compared with the QK method.

Main Results:

• Spearman's rank correlation between P-values from one- and two-stage methods ranged from 0.63 to 0.93.
• REML-based mixed-model association mapping showed better adherence to the nominal alpha-level and higher power for quantitative trait loci detection compared to the QK method.
• Two-step association mapping approaches did not substantially increase the empirical type I error rate compared to one-step approaches.

Conclusions:

• Mixed-model association mapping with REML-estimated kinship matrices is recommended for wheat.
• Both one-step and two-step association mapping approaches are viable, with two-step methods offering flexibility.
• These findings enhance the application of association mapping in autogamous species like wheat for genetic research.