Precision understanding of gene regulatory networks (GRN) is one of the major goals of modern quantitative and statistical genetics. Systems-level models contextualize GRNs providing a framework critical for insights into complex traits. Understanding complex disease requires that we understand the points in GRNs that are most susceptible to perturbation and how dysregulation within GRNs occurs. Questions of how GRNs may be compromised by environmental and genetic perturbations leading to disease are evolutionary questions of about robustness in the system. Are biological systems evolutionarily selected to be robust? Under what conditions is robustness violated? Answering these questions is a challenge we seek to address with this proposal. Genome wide association studies (GWAS) statistically connect genotypes to phenotypes, without explaining molecular interactions. Molecular biology directly ties gene function to phenotype through gene regulatory networks (GRNs), usually through the use of large effect (knock out /overexpression) alleles. The effect of polymorphisms among `wild type' alleles and how they impact the network are often unknown. GWAS and GRN approaches can be merged into a single framework, Structural Equation Modeling (SEM-GRN). This approach leverages the myriad of polymorphisms in natural populations to elucidate and quantitate the molecular pathways that underlie phenotypic variation. This framework can be used to evaluate GRN robustness. It is imperative that models of GRNs allow for a formal comparison between conditions and have the ability to predict the effect of allelic substitutions among a set of natural alleles. Once GRN modeling accounts for the effects of conditions it can be used to elucidate the relationships between GRN and phenotype variation. How individual alleles perturb the GRN, the regulatory components of GRNs; the degree to which GRNs are similar or different among conditions; and the identification of which alleles perturb the GRN in a condition specific manner lie at the heart of this proposal. The Drosophila sex determination (SD) GRN encapsulates all of these complexities. The SD- GRN is well studied with an established transcriptional regulatory cascade. There are known differences in the wiring of the GRN between males and females and between species. Within a sex/species `wild type' alleles have been categorized at several loci that have a quantitative effect on phenotype. Yet, there are still many regulatory inputs; downstream targets; and environmental effects that are unknown. We use this system to test and validate the novel SEM-GRN methods proposed to be developed here. We compare our novel approaches to eQTL based approaches and ensure broad applicability of the methods through extensive simulation and additional data analysis of the InR/Tor pathway in Drosophila and a reanalysis of the GTeX data in humans. All the methods here are directly relevant to natural populations including humans. We train future generation of researchers that will equally well tackle molecular quantitative genetic and statistical research and practice.
A gene regulatory network (GRNs) is usually modeled as a single entity regardless of the species, environment or genotype. Yet, we know that there is divergence in gene regulation that in different conditions and that dysregulation of GRNs leads to disease. Here we develop a condition specific GRN modeling approach. We will bring the power of Drosophila genetics to test and then validate the statistical models with further generalizability of the approach established using publically available human data.
