Using whole-genome sequence data to predict quantitative trait phenotypes in Drosophila melanogaster.

Area of Science:

• Genomics
• Quantitative Genetics
• Animal Breeding

Background:

• Predicting organismal phenotypes from genotype data is crucial for breeding, medicine, and evolutionary biology.
• Genomic prediction has primarily used SNP genotyping platforms, with limited application to complete genome sequences.

Purpose of the Study:

• To perform genomic prediction for starvation stress resistance and startle response in Drosophila melanogaster using extensive SNP data.
• To evaluate the predictive ability of the genomic best linear unbiased prediction (GBLUP) model with a large SNP dataset.

Main Methods:

• Sequencing the Drosophila Genetic Reference Panel to identify ~2.5 million single-nucleotide polymorphisms (SNPs).
• Constructing a genomic relationship matrix from SNP data and applying a GBLUP model.
• Assessing predictive ability via cross-validation, comparing GBLUP with BayesB and SNP selection strategies.

Main Results:

• Achieved predictive abilities of 0.239±0.008 for starvation resistance and 0.230±0.012 for startle response.
• The GBLUP model's predictive ability was comparable to the Bayesian method BayesB.
• Predictive ability decreased significantly only when using fewer than 150,000 SNPs.

Conclusions:

• Genomic prediction using a large number of SNPs is effective for complex traits in Drosophila.
• The predictive power likely arises from modeling subtle genetic structures via linkage disequilibrium, rather than population structure alone.
• These findings have implications for advancing genomic prediction strategies in diverse organisms.