Genotype–phenotype modeling considering intermediate level of biological variation: a case study involving sensory traits, metabolites and QTLs in ripe tomatoes

Abstract

Modeling genotype-phenotype relationships is a central objective in plant genetics and breeding. Commonly, variations in phenotypic traits are modeled directly in relation to variations at the DNA level, regardless of intermediate levels of biological variation. Here we present an integrative method for the simultaneous modeling of a set of multilevel phenotypic responses to variations at the DNA level. More specifically, for ripe tomato fruits, we use Gaussian graphical models and causal inference techniques to learn the dependencies of 24 sensory traits on 29 metabolites and the dependencies of those sensory and metabolic traits on 21 QTLs. The inferred dependency network which, though not essentially representing biological pathways, suggests how the effects of allele substitutions propagate through multilevel phenotypes. Such simultaneous study of the underlying genetic architecture and multifactorial interactions is expected to enhance the prediction and manipulation of complex traits.

Modelling of genotype by environment interaction and prediction of complex traits across multiple environments as a synthesis of crop growth modelling, genetics and statistics

“…Another common interaction occurs between flowering time and yield components, as reported by (Maurer et al, 2016;Mikolajczak et al, 2016;Wang et al, 2016). One alternative to model these interactions is to express explicitly the sensitivity of genes to environmental conditions via ecophysiological models, as shown for flowering time in barley by , or via graphical models and structural relations models (Wang & van Eeuwijk, 2014;Wang et al, 2015;Alimi, 2016). An alternative is the multi-locus HTL model we presented here, which is commonly used in human genetics, the so called SKAT method (sequence kernel association test (Wu et al, 2011)).…”

“…However, a more elegant and complete representation of the relationship between QTLs, APSIM parameters, intermediate and target traits can be achieved with a network representation (Neto et al, 2010;Alimi, 2016). Best, if multiple layers of information (QTLs, APSIM parameters and final traits) are modelled in a multi-level network (Wang & van Eeuwijk, 2014;Wang et al, 2015).…”

“…The prediction of phenotypes can be improved by modelling intermediate levels of biological variation in between the DNA level (SNPs and sequence information) and the final phenotype (target trait): gene expression, proteins, metabolites, methylation, etc.. Network models are a popular type of model for combining the variation of different types of traits at multiple levels of biological organization, including target phenotypic traits at the highest level (Welch et al, 2003(Welch et al, , 2005Neto et al, 2008Neto et al, , 2010Scutari et al, 2014;Wang and van Eeuwijk, 2014;Wang et al, 2015). Network models show the behavior of multiple traits in their dependence on each other.…”

“…If the aim is to characterize the genetic basis for multiple levels of phenotypic organization, multi-level directed networks are necessary. For example, Wang and van Eeuwijk (2014) and in Wang et al (2015) used multi-level directed networks to characterize the relationships between QTLs, metabolites and sensory traits in tomato.…”

Computational aspects underlying genome to phenome analysis in plants

“…The QTL + phenotype supervised orientation (QPSO) approach, developed in the van Eeuwijk laboratory (Wang and van Eeuwijk, ; Wang et al ., ), aims to generate directed networks between phenotypic traits by using known sparse QTL to orient the network, extending earlier work on gene network reconstruction and cleverly combining different data domains.…”