Connecting Phenotypes or Traits to Biological Processes and Molecular Functions in Plants

Current studies of the genetic basis of phenotypes or traits do not provide direct access to the molecular and/or physiological basis of such traits. We here present a large-scale inventory of connections between phenotypes/traits and gene functions (both biological processes and molecular functions as defined by Gene Ontology), based on the integration of QTL and GWAS data with gene function predictions. We assess the overrepresentation of predicted gene functions  in regions associated with traits in several plant QTL/GWAS datasets. For the prediction of biological processes, our network-based method offers a considerable added value (Bargsten et al., submitted). Compared to  sequence only-based prediction strategies, the number of statistically significant links between biological processes and traits was much higher. Various approaches, including text mining,  provided validation for the predictions. The connections predicted between traits/phenotypes and gene function help to unravel the underlying (molecular) biology of such traits and may define targets for breeding. For example, in a rice dataset, predictions include strong associations between (a) the root number and terpene biosynthetic processes as well (b) leaf senescence and starch metabolism, and (c) potassium uptake and glycogen/glucan related processes. In Arabidopsis, several traits related to flowering were predicted to be connected to defense-related processes. Comparison with QTL fine-mapping results  show this is a promising approach to prioritize putative candidate genes underlying traits-of-interest.