2012 Annual Report
1a.Objectives (from AD-416):
The specific objectives of this agreement are to Identify differentially expressed (DE) genes in blood in response to PRRSV infection; Determine putative gene sets and pathways that predict a pig's ability to clear PRRSV infection and maintain weight gain; and Validate utility of gene sets and pathways for prediction of responsiveness to PRRSV infections in multiple populations. Predictive blood tests of pigs with improved PRRS disease resistance and growth maintenance; increased understanding of mechanisms involved in pig responses to PRRSV infection; scientific publications.
1b.Approach (from AD-416):
The university's Data storage will be centralized at ISU via the PRRS Host Genetics Consortium (PHGC) database and for this project the microarray data through ANEXdb. A tool will be developed to accept two-color array data as well as updated annotations for the Pigoligoarray oligonucleotides. For the first objective the microarray data analysis, lists of important DE genes will be annotated using DAVID tools, clustered and drawn into networks to generate informative networks that describe the common response to PRRSV over time in high viral load and low viral load animals. These networks will be further reinforced and expanded using orthogonal data such as cytokine expression and the second objective data. Will overlay the first and second objective networks to potentially clarify modules and hubs responding in vivo, and attempt to understand regulatory networks across datasets. For Classifier development will use all the first and second objective expression data (using gene sets clustered by expression pattern within and across Objectives., separately or in combination with QPCR, cytokine or other phenotypic data) to create and evaluate several classifiers using machine learning algorithms implemented in Weka. Initially, 24-48 genes will be used from each sample using sets of RNAs from PHGC samples representing the 4 different virus/weight categories to construct the classifiers. The performance of the resulting classifiers (as measured by classification accuracy, sensitivity, specificity, and area under ROC curve), will be evaluated using rigorous statistical cross-validation. Genes predicted to be good classifiers will be tested by coPIs using QPCR on additional samples for the iterative refinement process to improve classifier predictions.
This functional genomics project uses samples collected from pigs infected with porcine reproductive and respiratory syndrome virus (PRRSV), a major swine pathogen causing losses to the U.S. pig industry of $664 million per year. The goal is to determine which anti-viral response pathways differ in PRRS-resistant versus PRRS-susceptible pigs using samples collected as part of the PRRS Host Genetics Consortium (PHGC). ARS Researchers at Beltsville, Maryland (BARC) have partnered with Iowa State University (ISU) scientists to analyze blood gene expression data generated by the project. Data storage has been centralized at ISU via the secure, password protected PHGC database (www.animalgenome.org/lunney/). Specifically, the microarray data is being archived using an integrated animal anotation and microarray expression database (ANEXdb), which has been expanded to accommodate this project's two-color gene expression data. Michigan State University (MSU) scientists used Pigoligoarrays to evaluate changes in gene expression of blood RNA from 12 pigs collected at 0, 4, 7, 11, 14, 28, and 42 days post infection (dpi). Biostatistical and bioinformatic analyses were performed at MSU interacting with ISU scientists. Sophisticated statistical tools were used to generate informative gene networks that differentiate the PRRSV response patterns in pigs representing different virus/weight categories. Those analyses have resulted in the decision to focus on 0, 4, and 7 dpi samples so that gene expression of early anti-PRRSV infection can be evaluated in a more robust statistical manner. Targeted quantitative polymerase chain reaction assays will help to affirm which genes are correlated with viral load and/or weight gain by comparing the (1) most desirable, PRRS-resistant, low virus, high weight gain pigs with the (2) worst, PRRS-susceptible, high virus, low weight gain pigs. As this work progresses more systems biology based analyses will be performed at ISU to develop predictive gene expression pathways and classifier genes that identify pigs which resist PRRSV infection and grow normally.