Performing Department
Veterinary Population Medicine
Non Technical Summary
Billions of dollars have been invested annually in research aimed at trying to determine the genetic contribution to complex traits where both genetics and the environment impact the severity of the phenotype. This major investment was built around the promise of using an individual's unique genetic code in order to make more precise management or medical decisions. However, despite decades of research and extensive collaborative projects focused on connecting genetics to complex traits, the implementation of precision management and medicine have not yet come to fruition. The recent USDA Blueprint identified that the continued development of genome tools and resources is necessary to advance genome to phenome discovery. Major steps towards this movement are integration of datasets that incorporate multiple levels of 'omics data (i.e. genetics, metabolomics, and gene transcription and regulation), as well as clearly and consistently describe complex traits across species.The first objective of this project is to create a streamlined workflows for researchers to be able to integrate multi-omics datasets across species. This objective will be accomplished by leveraging data collected from the equine metabolic syndrome (EMS) model as proof of concept. Data and statistical analyses will be performed to assess correlations between (1) the genome, metabolome and phenome by performing a genome-wide association analysis and co-mapping regions of the genome, (2) genome, transcriptome and phenome by performing differential gene expression and gene set enrichment analysis, and (3) genome, transcriptome and metabolome through co-expression network analysis. This will enable development of workflows and procedure documentation that can be broadly applied to other animal systems. The second objective of this project is to develop ontology-based tools that facilitate genotype to phenotype discovery through consistent phenotype descriptors. This objective will be accomplished by curating important traits to the USDA agricultural community from phenotype ontology databases, adding key missing descriptors, and creating a gene to phenotype database using similarities across species. This work will be developed into an open-source phenotype-based ontology tools for use in agricultural species.
Animal Health Component
100%
Research Effort Categories
Basic
50%
Applied
50%
Developmental
(N/A)
Goals / Objectives
The major goals of this project are to develop infrastructure that enables integration and translation of multi-omics data into applied knowledge for animal health, performance and production.Aim 1 of the proposal focuses on enabling integration and translation of multi-omics data. The objectives for this aim which serve to accomplish this goal are (1) integration of the metabolome, phenome and genome, (2) integration of the transcriptome, phenome and genome, (3) co-expression network analysis transcriptome, metabolome and phenome, (4) development of an integrative genetics workflow with multiple analytic platforms.Aim 2 of this proposal focuses on applied knowledge for animal health, performance and production. The objectives in aim 2 which server to accomplish this major goal are (1) development of consistent phenotype descriptors in agricultural species, (2) applying phenotypic descriptors across multiple mammalian species, and (3) development of resource tools for linking genetic data to phenotype.The workflows and tools developed from this proposal will be open-source, extendible to any species and phenotype, and will substantially facilitate the movement from genome to phenome in animal genetic studies. Improved rates of genetic allele discovery for complex traits will finally allow the implementation of precision management, medicine, and breeding to come to fruition.
Project Methods
Aim 1: Enabling the integration of multi-omics data to link genotype to phenotype. My overall approach is to show proof of concept using data from EMS to assess correlations between (1) the genome, metabolome and phenome, (2) genome, transcriptome and phenome, and (3) genome, transcriptome and metabolome. This will enable development of workflows and procedure documentation that can be broadly applied to other animal systems. Workflows will use open source software and be made available as Docker containers that can be deployed independently.Relationships between the genome, metabolome and phenome: Metabolomic data from 286 Morgans and 264 Welsh ponies will be available at the start of this fellowship and used to determine which metabolites are statistically associated with EMS clinical traits. Analyses will be performed using the R software package NLME with EMS traits as quantitative response variables, relative abundance of the metabolites as predictor variables, age and sex as fixed effects, and environment as a random effect. Metabolites that were statistically associated with one or more EMS traits will be identified via mass spectrometry at the UMN Center for Proteomics and Mass Spectrometry. Genetic risk loci will be identified by within-breed genome-wide association. Statistically significant regions on GWA will be assessed to determine if they co-map with within-breed GWA regions identified previously for EMS traits. Co-mapped regions will be interrogated using imputed whole genome sequencing (previously calculated at a >98.9% imputation concordance from 1.2M SNPs using Beagle4) and haplotype analysis for genetic variant and candidate gene discovery.Relationships between the transcriptome, phenome, and genome: Gene expression in muscle and adipose tissue has been quantified by 20M RNA-seq reads from 21 Morgans and 20 Welsh ponies phenotyped for EMS. I will identify differential expressed genes between EMS clinical traits. Groups of genes significantly correlated to an EMS clinical trait or gene sets based on prior knowledge will be tested jointly for association using Gene Set Enrichment Analysis. Regression-based mediation analysis will be used to assess the effect of the SNP GWA regions of interest and gene expression data.Relationship between the genome, transcriptome and metabolome: Evaluation of metabolomic and transcriptomic co-expression networks will be performed using Camoco to infer functional components of network interactions. Network locality will be used to identify positional candidate genes that are proportionally more connected to genes locally (candidate genes from GWA) than they are globally (all genes in the network).Aim 2. Developing resources to apply standardized phenotyping to clinical applications. This objective addresses the need for veterinarians and producers to apply knowledge gained from multi-omics data directly to the animal. By carefully and consistently describing complex traits, we will be able to better understand the contribution of genetics to phenotypes, and to leverage cross-species information to gain insight about complex phenotypes. The expected outcome for this objective is an open-source tool for linking trait-relevant omics data to causal genes.Developing consistent phenotype descriptors: Initially we will use the set of traits and disorders represented in OMIA, linking the curated phenotypic information (e.g., symptoms and clinical signs) to terms in the phenotype ontologies. We anticipate that the OMIA dataset will provide a core set of well-defined traits, including Mendelian disorders that are well studied in other species. Moreover, this approach helps to preserve the curated OMIA data by ensuring that it is disseminated to additional online resources and enables us to work with the existing phenotype ontology community to ensure that we are using community agreed upon best practices. Next, we will extend this core set to include phenotypes important for the USDA livestock research community. Traits that are well studied by the livestock community are represented by the QTLDb and include observations relating to health, growth, performance, reproduction and behavior. We will prioritize annotation of these traits by surveying the research community to identify the traits should be curated. This survey will be distributed through AngenMap and the Plant and Animal Genome Workshop participants.Applying phenotypic descriptors across multiple mammalian species: While USDA animals have amassed a significant body of information about genetic markers and heritability, mammalian biomedical models have a wealth of information about how genes contribute to phenotype. I will integrate human, rodent and livestock phenotype data using orthologs to apply phenotypic descriptors across multiple mammalian species. I will identify strictly orthologous genes (1:1) using standardized resources such as the HGNC Comparative Orthologs Predictions (HCOP) tool, which collates ortholog predictions from multiple sources (including Ensembl Compara, Homologene, OrthoDb, etc). This will provide a core set of orthologs that can be used to transfer additional predicted phenotype information from human and rodent genes to livestock genes. Human and mouse phenotype information will be obtained from the Human Phenotype ontology project, the Mouse Genome Database and Mouse Pheome Database resources, and transferred phenotypes will be recorded as 'inferred by orthology' following ontology community best practice standards. This will provide a first-pass core set of genes with phenotype descriptors that are directly comparable across species and to existing trait resources such as QTLDb.Providing resources for linking genetic data to phenotypes: I will develop a freely available, publicly accessible tool to ensure that USDA researchers are able to utilize phenotype data from this proposal. For each USDA livestock species, genes and traits will be integrated by applying the phenotypic descriptors to genomics coordinates. I will work with tool developers at AgBase, QTLDb and CyVerse to develop software that accepts genomic regions of interest and returns phenotype descriptors associated with these regions. I anticipate preliminary testing of this tool using EMS data from Aim 1. This will be followed by further testing, feedback and refinement by recruiting researchers from the USDA livestock community who have their own genetic data sets. As a part of this tool development, I will provide documentation of phenotype association processes, completed examples, and online training materials. Dr. McCarthy has experience at developing and providing online analysis tools, and we anticipate an iterative process based upon community feedback and as other data types become available. For example, we anticipate that the Functional Annotation of Animal Genomes project will result in additional genomic annotation and information about regulatory regions that is linked with phenotype data and will develop this tool to incorporate this data when it becomes available.Milestones and Evaluations:Validation of workflow with canine metabolomics and genomics dataset from collaborators and freely available datasets in bovine transcriptomicsPublications: Equine serum metabolomics GWA publication, equine transcriptomics and network analysis, phenotype ontologies in agricultural speciesWorkflow publication and documentation on GitHub. Documentation will be peer reviewed by outside experts through the Mozilla mentorship programPresentation of the integration of multi-omics data for EMS and final workflowDistribution of survey to AngenMap and the Plant and Animal Genome Workshop participantsPresentation of phenotype ontology data at bioinformatics conferenceEvaluation of documentation and usability through MozillaRelease of open-source toolWorkshop on tool usage at Havemeyer Horse Genome Workshop in 2022