G2P

G2P provides an integrative environment for genotype-to-phenotype prediction to support genomic selection (GS)-assisted breeding in the seed industry.

Key Features:

• Comprehensive Model Library: Includes a library of 16 state-of-the-art genomic selection models for comparative analysis.
• Evaluation Metrics: Implements 13 evaluation metrics to enable unbiased assessment of GS model performance.
• Parallel Processing Capability: Optimized for high-performance computing clusters to enable parallel execution of multiple models.
• Auto-Ensemble Algorithms: Employs auto-ensemble algorithms that automatically select the most precise models or integrate results from multiple models based on evaluation outcomes.
• Training Set Refinement: Refines training sets using genetic diversity analysis to maintain prediction precision with fewer samples, reducing phenotyping costs.

Scientific Applications:

• Genomic selection-assisted breeding: Supports model comparison and selection for GS-assisted breeding programs in the seed industry.
• Cross-species and multi-trait prediction: Addresses variability in prediction precision across different species and traits.
• Crop variety improvement: Facilitates development of improved crop varieties with desirable phenotypic characteristics.

Methodology:

Combines model evaluation, model selection, and ensemble techniques within a unified framework, performs training set refinement via genetic diversity analysis, and supports parallel execution on high-performance computing clusters.