To what extent gene connectivity within co-expression network matters for phenotype prediction?

Abstract

Recent literature on the differential role of genes within networks distinguishes core from peripheral genes. If previous works have shown contrasting features between them, whether such categorization matters for phenotype prediction remains to be studied. We sequenced RNA in a Populus nigra collection and built co-expression networks to define core and peripheral genes. We found that cores were more differentiated between populations than peripherals while being less variable, suggesting that they have been constrained through potentially divergent selection. We also showed that while cores were overrepresented in a subset of genes deemed important for trait prediction, they did not systematically predict better than peripherals or even random genes. Our work is the first attempt to assess the importance of co-expression network connectivity in phenotype prediction. While highly connected core genes appear to be important, they do not bear enough information to systematically predict better quantitative traits than other gene sets.Gene-to-gene interaction is a pervasive although elu-2 sive phenomenon underlying phenotype expression. 3Genes operate within networks with more or less me-4 diated actions on the phenome. Systems biology ap-5 proaches are required to grasp the functional topol-6 ogy of these networks and ultimately gain insights 7 into how gene interactions interplay at different bio-8 logical levels to produce global phenotypes (Mackay 9 et al., 2009). New sources of information and their 10 subsequent use in the inference of gene networks are 11 populating the wide gap existing between phenotypes 12 and DNA sequences and, therefore, opening the door 13 to systems biology approaches for the development 14 of context-dependent phenotypic predictions. RNA 15 98 Montanucci et al., 2011). On the other hand, genes 99 being better predictors with an interactive model are 100supposed to be upstream in pathways. We expect pe-101 ripheral genes to behave interactively, with a lower 102 but relatively more variable expression level. With a 103 lower variation, we also expect core genes to be worse 104 predictors for traits than peripheral genes unless the 105 former also bear larger effects. 106 To answer the questions concerning the respective 107 roles of core and peripheral genes on phenotypic vari-108 ation, we have sequenced the RNA of 459 samples of 109 black poplar (Populus nigra), corresponding to 241 110 genotypes, from 11 populations representing the nat-111 ural distribution of the species across Western Eu-112 rope. We also have for each of these trees phenotypic 113 records for 17 traits, covering the growth, phenol-114 ogy, physical and chemical properties of wood. They 115 cover two different environments where the trees were 116 grown in common gardens, in central France and 117 northern Italy. With the transcriptomic data, we 118 built a co-expression network in order to define con-119 trasting gene sets according to their connectivity 120 within the network. We then asked whether these 121 c...

Polygenic adaptation and negative selection across traits, years and environments in a long-lived plant species (Pinus pinasterAit., Pinaceae)

“…implies that virtually all the expressed genes in a relevant tissue likely influence the functions 126 of core genes that are directly relate with the phenotype. In this regard, recent work with 127 expression networks revealed that genes with higher connectivity experienced higher 128 intensities of selection and consequently presented higher variation, increasing the probability 129 of being detected by traditional GWAS (Mähler et al 2017; Chateigner et al 2019). In 130 addition, the study of polygenic adaptation at pathway level revealed that gene sets related 131 with resistance to pathogens in humans showed enriched signals of selection (Daub et al 132 2013), highlighting its importance in shaping human adaptation.…”