Codominant scoring of AFLP in association panels

Area of Science:

• Genetics
• Bioinformatics
• Plant Breeding

Background:

• Amplified Fragment Length Polymorphism (AFLP) markers are widely used in genetic studies.
• Accurate codominant scoring of AFLP markers is crucial for association mapping.
• Existing methods may lack precision without prior genotype probability knowledge.

Purpose of the Study:

• To optimize and illustrate codominant scoring of AFLP markers using normal mixture models.
• To enhance the performance of the EM-algorithm for AFLP data analysis.
• To provide accurate posterior genotype probabilities for association mapping.

Main Methods:

• Fitting normal mixture models to AFLP band intensities.
• Utilizing the EM-algorithm with optimized parameter initialization and restrictions.
• Applying data transformations (e.g., square root) and outlier removal.
• Developing diagnostic tools for AFLP marker data quality.

Main Results:

• Empirical evidence supports the square root transformation for band intensity.
• The method generates posterior genotype probabilities superior to standard scoring categories.
• Developed R software facilitates normal mixture modeling and visualization for AFLP data.
• Applied methodology to a tomato association panel of 94 hybrids with 1,175 markers.

Conclusions:

• The optimized normal mixture model approach provides reliable codominant scoring for AFLP markers.
• Posterior genotype probabilities enhance the accuracy of association mapping.
• The developed software and diagnostics improve AFLP data quality assessment and analysis.