Jicai Jiang

The overall goal of this integrated proposal is to develop a functional genomics approach to identify genomic and epigenomic biomarkers and enable selective breeding of sows that are at a minimal risk of producing offspring that display the negative IUHS phenotypes (Fig. 1). The genomic selection models will be biologically validated using complementary measures of offspring behavior and physiology in response to production stressors, immune challenges, heat stress exposure, and the determination of transgenerational effects on the pig epigenome. The success of genomic selection for sows that are at a minimal risk of producing offspring that display the negative IUHS phenotypes depends on the availability of phenotypes that are heritable and repeatable, can be measured on a large number of animals, and that are representative of the postnatal consequences of IUHS in swine offspring.

The objective of this project is to identify opportunities to optimize swine production systems for efficiency of nutrient utilization spanning the spectrum of variability generated by the pig microbiome interactions. The existence of genetic control over the abundance of particular taxa and the link of these with energy balance and growth has been shown in model organisms. Little evidence has been so far presented in pigs. The gut microbiome is an essential component of variability of growth in several species. Swine do not escape this rule and the identification of the significance of gut microbiota in the growth process will be a game changer in the achievement of sustainable and efficient lean meat production. Within this proposal we will: Characterize the interaction between host and longitudinal microbiome development in swine and its effect on efficient growth, Model the genomic of longitudinal variability in microbiota development, Link the fecal and cecal luminal metagenomic communities in pigs, Integrate data generated by longitudinal analysis with already generated evidence. Making microbiome composition part of the breeding objective or directly manipulating the microbiome in populations under selection through diet or other artificial means could revolutionize the swine industry. We will identify efficient individuals at maintaining a favorable microbiome. We hypothesize that a portion of this ability is under genetic control making these individuals better able to cope with changes in dietary and environmental conditions. We propose an innovative approach to develop new selection methods using a combination of genomics and microbiome.

Though genomic selection has been successful in accelerating genetic improvement, we still know very little about the genetic architecture of quantitative traits. This poses a challenge for researchers to develop a model that can achieve the full potential of genomic evaluation. Recent progress on functional annotation of animal genomes (FAANG) provides an opportunity to address the challenge in hopes that the functional information can help to decipher the genotype-to-phenotype relationship; however, there is a lack of methods for effectively using a variety of functional annotations in quantitative genetic analysis. The goal of this proposal is to develop a unified mixed-model method for integrating functional annotations into genome-to-phenome analysis. Our method can 1) reveal the genetic architecture of quantitative traits by jointly quantifying the contributions of various functional annotations to quantitative trait variation, 2) identify trait-relevant tissues or cell-types by the analysis of tissue-/cell-type-specific functional annotations, and 3) improve genomic predictions by incorporating functional annotations. The project will be addressing three main objectives, aimed at 1) developing a unified method for incorporating numerous functional annotations simultaneously in mixed-model genetic analysis, 2) implementing the method in an efficient software tool, and 3) demonstrating the method with dairy cattle data sets. This project will provide one of the first methods for integrating FAANG data into various genetic analyses and present several demonstrations of its use.

While there is evidence of genetic variability for heat stress response in swine, there is a lack of (bio)markers that could allow the understanding and implementation of breeding programs aimed at selecting for a more heat tolerant sow. The use of the gut microbial flora as prognostic aid has been long investigated in humans and microbiota profiles related to heat stress could serve this function. The goal of this proposal is to develop an enhanced selection tool, including the most effective combination of gut microbiome and host genomic/metabolomic/phenomic features. We will extract these features from in-depth sow characterizations in research settings and translate them to a large-scale dataset collected at the commercial level. The project will be addressing two main objectives, aimed at 1) understanding how host genomic, metabolomic, and gut microbial features regulate sow tolerance to heat stress and their potential use as biomarkers and 2) developing microbiome-enhanced genomic prediction tool itself. The NCSU group (Dr. Tiezzi, Dr. Maltecca) is currently one of the most active in the USA working on the genomics of heat tolerance in swine and on the use of gut microbiome information in breeding programs. This project significantly expands the relevance of a current USDA-sponsored project (award number 2020-67015-31575), awarded to a part of the proposing investigators, adding gut microbiome, host transcriptomics and host metabolomics information. This project will provide the first omni-comprehensive characterization through unprecedented in-depth phenotyping. The project will also provide much-needed selection tools to exploit genomic selection for heat-tolerance in swine.������������������������������������������������������������������������������������������

We aim to develop and maintain a software package for large-scale genome-wide association studies and fine-mapping (BFMAP) in livestock populations. It can efficiently decipher functional annotation data and incorporate the data into fine-mapping. BFMAP will be useful for understanding genetic architecture of complex traits in livestock.

Genomic prediction has become a hallmark of the use of genomics to enhance genetic selection in livestock, including pigs, by increasing the accuracy of selection. This is in particular important for disease resilience, which cannot be evaluated directly in the high-health nucleus breeding herds where most selection occurs. Several studies have shown that increasing the number of SNPs that animals in the training data are genotyped for has limited impact on the accuracy of selection, because of the high linkage disequilibrium. This is even the case when animals are imputed up to sequence, although in that case, most causative mutations are expected to be included in the genotype data that is used for training. The problem, however, is that the genomic prediction model is not provided information to differentiate causal SNPs from SNPs that are in high LD with them. This can be addressed by providing the genomic prediction model functional genomic information by classifying SNPs based on functional genomic information and to allow the genomic prediction model to put differential weighting on the resulting SNP classes. To enable this, co-PI Jicai Jiang has developed a fast variational Bayes method of genomic prediction that allows multiple SNP classes to be fitted. Other methods to accommodate functional genomic information are available also. Thus, the objective of this proposal is to investigate the impact of the integration of various types of functional genomic information on the accuracy of genomic prediction for disease resilience and related traits in grow-finish pigs. For this purpose, we will capitalize on a data set of over 5,000 Landrace x Yorkshire barrows that were subjected to a natural polymicrobial disease challenge in the late nursery stage at a research facility in Quebec, Canada. This data set includes data on pigs from 7 breeding companies that are members of the PigGen Canada research consortium. Phenotypes collected during the challenge include mortality, health treatments, subjective health scores, growth rate, feed and water intake and behavior, and carcass traits. These phenotypes will be the targets for genomic prediction. The proposed work includes the following activities: 1) Impute all pigs from the current ~480K SNP genotypes that are available on all pigs up to sequence. For this purpose, we will use the Swine Imputation (SWIM) server developed and managed by co-PI Wen Huang. The SWIM project (https://quantgenet.msu.edu/swim/index.html) was supported in part by USDA Swine Genome Coordinator funds. 2) Classify SNPs based on functional genomic information. Several classifications will be developed, including their location and potential function in the genome and immune response, chromatin states, etc. We will also use blood transcriptome data that we have available on over 2500 of these pigs prior to their entry in the disease challenge and associations of this transcriptome with disease resilience. 3) Implement genomic prediction analyses that utilize the above classification information, including variational Bayes methods. Accuracy of the resulting genomic predictions will be evaluated using cross-validation approaches.