Partial correlation analysis of transcriptomes helps detangle the growth and defense network in spruce

Abstract

O_LIIn plants, there can be a trade-off between resource allocations to growth versus defense. Here, we use partial correlation analysis of gene expression to make inferences about the nature of this interaction.\nC_LIO_LIWe studied segregating progenies of Interior spruce subject to weevil attack. In a controlled experiment, we measured pre-attack plant growth and post-attack damage with several morphological measures, and profiled transcriptomes of 188 progeny.\nC_LIO_LIWe used partial correlations of individual transcripts (ESTs) with pairs of growth/defense traits to identify important nodes and edges in the inferred underlying gene network, e.g., those pairs of growth/defense traits with high mutual correlation with a single EST transcript. We give a method to identify such ESTs.\nC_LIO_LIA terpenoid ABC transporter gene showed strongest correlations (P=0.019); its transcript represented a hub within the compact 166-member gene-gene interaction network (P=0.004) of the negative genetic correlations between growth and subsequent pest attack. A small 21-member interaction network (P=0.004) represented the uncovered positive correlations.\nC_LIO_LIOur study demonstrates partial correlation analysis identifies important gene networks underlying growth and susceptibility to the weevil in spruce. In particular, we found transcripts that strongly modify the trade-off between growth and defense, and allow identification of networks more central to the trade-off.\nC_LI