Applying network and genetic analysis to the potato metabolome

Abstract

Compositional traits in potato [Solanum tuberosum L.] are economically important but genetically complex, often controlled by many loci of small effect; new methods need to be developed to accelerate analysis and improvement of such traits, like chip quality. In this study, we used network analysis to organize hundreds of metabolic features detected by mass spectrometry into groups, as a precursor to genetic analysis. 981 features were condensed into 44 modules; module eigenvalues were used for genetic mapping and correlation analysis with phenotype data collected by the Solanaceae Coordinated Agricultural Project. Half of the modules were associated with at least one SNP according to GWAS; 11 of those modules were also significantly correlated with chip color. Within those modules features associated with chipping provide potential targets for selection in addition to selection for reduced glucose. Loci associated with module eigenvalues were not evenly distributed throughout the genome but were instead clustered on chromosomes 3, 7, and 8. Comparison of GWAS on single features and modules of clustered features often identified the same SNPs. However, features with related chemistries (for example, glycoalkaloids with precursor/product relationships) were not found to be near neighbors in the network analysis and did not share common SNPs from GWAS. Instead, the features within modules were often structurally disparate, suggesting that linkage disequilibrium complicates network analyses in potato. This result is consistent with recent genomic studies of potato showing that chromosomal rearrangements that create barriers to recombination are common in cultivated germplasm.