Needles: Toward Large-Scale Genomic Prediction with Marker-by-Environment Interaction.

Area of Science:

• Agricultural Science
• Genetics
• Computational Biology

Background:

• Genomic prediction uses genotypic markers to forecast crop performance.
• Marker effects vary with environmental conditions, necessitating analysis of marker-by-environment interactions.
• Analyzing large-scale trial data with these interactions demands significant computational resources.

Purpose of the Study:

• To present a high-performance computing solution, Needles, for analyzing large-scale genomic trial data with marker-by-environment interactions.
• To demonstrate the efficiency and effectiveness of Needles in handling complex genomic prediction models.
• To explore the impact of environmental data on the accuracy of genomic prediction.

Main Methods:

• Development of a high-performance computing solution (Needles) utilizing sparse matrix formalism and distributed computing.
• Tailoring the computational framework to the algebraic properties of genomic prediction models.
• Analysis of simulated trial data to evaluate prediction accuracy and identify quantitative trait loci (QTL).

Main Results:

• Needles enables large-scale genomic prediction analyses with marker-by-environment interactions in reasonable time frames.
• Increased records per environment in training data improve the prediction accuracy of markers with high environmental interaction.
• The framework facilitates the discovery of QTL with significant contributions in specific environmental conditions.

Conclusions:

• The Needles framework provides an efficient computational solution for genomic prediction incorporating marker-by-environment interactions.
• Accurate prediction of crop performance in specific environments is achievable with sufficient data and advanced computational tools.
• This approach supports plant breeders in selecting hybrid lines with optimized agronomic performance for targeted environments.