Adjusting for gene-specific covariates to improve RNA-seq analysis.

SUMMARY: This article suggests a novel positive false discovery rate (pFDR) controlling method for testing gene-specific hypotheses using a gene-specific covariate variable, such as gene length. We suppose the null probability depends on the covariate variable. In this context, we propose a rejection rule that accounts for heterogeneity among tests by using two distinct types of null probabilities. We establish a pFDR estimator for a given rejection rule by following Storey's q-value framework. A condition on a type 1 error posterior probability is provided that equivalently characterizes our rejection rule. We also present a suitable procedure for selecting a tuning parameter through cross-validation that maximizes the expected number of hypotheses declared significant. A simulation study demonstrates that our method is comparable to or better than existing methods across realistic scenarios. In data analysis, we find support for our method's premise that the null probability varies with a gene-specific covariate variable. AVAILABILITY AND IMPLEMENTATION: The source code repository is publicly available at https://github.com/hsjeon1217/conditional_method.

Gene expression in tonsils in swine following infection with porcine reproductive and respiratory syndrome virus.

BACKGROUND: Porcine reproductive and respiratory syndrome (PRRS) is a threat to pig production worldwide. Our objective was to understand mechanisms of persistence of PRRS virus (PRRSV) in tonsil. Transcriptome data from tonsil samples collected at 42 days post infection (dpi) were generated by RNA-seq and NanoString on 51 pigs that were selected to contrast the two PRRSV isolates used, NVSL and KS06, high and low tonsil viral level at 42 dpi, and the favorable and unfavorable genotypes at a genetic marker (WUR) for the putative PRRSV resistance gene GBP5. RESULTS: The number of differentially expressed genes (DEGs) differed markedly between models with and without accounting for cell-type enrichments (CE) in the samples that were predicted from the RNA-seq data. This indicates that differences in cell composition in tissues that consist of multiple cell types, such as tonsil, can have a large impact on observed differences in gene expression. Based on both the NanoString and the RNA-seq data, KS06-infected pigs showed greater activation, or less inhibition, of immune response in tonsils at 42 dpi than NVSL-infected pigs, with and without accounting for CE. This suggests that the NVSL virus may be better than the KS06 virus at evading host immune response and persists in tonsils by weakening, or preventing, host immune responses. Pigs with high viral levels showed larger CE of immune cells than low viral level pigs, potentially to trigger stronger immune responses. Presence of high tonsil virus was associated with a stronger immune response, especially innate immune response through interferon signaling, but these differences were not significant when accounting for CE. Genotype at WUR was associated with different effects on immune response in tonsils of pigs during the persistence stage, depending on viral isolate and tonsil viral level. CONCLUSIONS: Results of this study provide insights into the effects of PRRSV isolate, tonsil viral level, and WUR genotype on host immune response and into potential mechanisms of PRRSV persistence in tonsils that could be targeted to improve strategies to reduce viral rebreaks. Finally, to understand transcriptome responses in tissues that consist of multiple cell types, it is important to consider differences in cell composition.

rmRNAseq: differential expression analysis for repeated-measures RNA-seq data.

MOTIVATION: With the reduction in price of next-generation sequencing technologies, gene expression profiling using RNA-seq has increased the scope of sequencing experiments to include more complex designs, such as designs involving repeated measures. In such designs, RNA samples are extracted from each experimental unit at multiple time points. The read counts that result from RNA sequencing of the samples extracted from the same experimental unit tend to be temporally correlated. Although there are many methods for RNA-seq differential expression analysis, existing methods do not properly account for within-unit correlations that arise in repeated-measures designs. RESULTS: We address this shortcoming by using normalized log-transformed counts and associated precision weights in a general linear model pipeline with continuous autoregressive structure to account for the correlation among observations within each experimental unit. We then utilize parametric bootstrap to conduct differential expression inference. Simulation studies show the advantages of our method over alternatives that do not account for the correlation among observations within experimental units. AVAILABILITY AND IMPLEMENTATION: We provide an R package rmRNAseq implementing our proposed method (function TC_CAR1) at https://cran.r-project.org/web/packages/rmRNAseq/index.html. Reproducible R codes for data analysis and simulation are available at https://github.com/ntyet/rmRNAseq/tree/master/simulation.

Acute systemic inflammatory response to lipopolysaccharide stimulation in pigs divergently selected for residual feed intake.

BACKGROUND: It is unclear whether improving feed efficiency by selection for low residual feed intake (RFI) compromises pigs' immunocompetence. Here, we aimed at investigating whether pig lines divergently selected for RFI had different inflammatory responses to lipopolysaccharide (LPS) exposure, regarding to clinical presentations and transcriptomic changes in peripheral blood cells. RESULTS: LPS injection induced acute systemic inflammation in both the low-RFI and high-RFI line (n = 8 per line). At 4 h post injection (hpi), the low-RFI line had a significantly lower (p = 0.0075) mean rectal temperature compared to the high-RFI line. However, no significant differences in complete blood count or levels of several plasma cytokines were detected between the two lines. Profiling blood transcriptomes at 0, 2, 6, and 24 hpi by RNA-sequencing revealed that LPS induced dramatic transcriptional changes, with 6296 genes differentially expressed at at least one time point post injection relative to baseline in at least one line (n = 4 per line) (|log2(fold change)| ≥ log2(1.2); q < 0.05). Furthermore, applying the same cutoffs, we detected 334 genes differentially expressed between the two lines at at least one time point, including 33 genes differentially expressed between the two lines at baseline. But no significant line-by-time interaction effects were detected. Genes involved in protein translation, defense response, immune response, and signaling were enriched in different co-expression clusters of genes responsive to LPS stimulation. The two lines were largely similar in their peripheral blood transcriptomic responses to LPS stimulation at the pathway level, although the low-RFI line had a slightly lower level of inflammatory response than the high-RFI line from 2 to 6 hpi and a slightly higher level of inflammatory response than the high-RFI line at 24 hpi. CONCLUSIONS: The pig lines divergently selected for RFI had a largely similar response to LPS stimulation. However, the low-RFI line had a relatively lower-level, but longer-lasting, inflammatory response compared to the high-RFI line. Our results suggest selection for feed efficient pigs does not significantly compromise a pig's acute systemic inflammatory response to LPS, although slight differences in intensity and duration may occur.

Post-weaning blood transcriptomic differences between Yorkshire pigs divergently selected for residual feed intake.

BACKGROUND: Improving feed efficiency (FE) of pigs by genetic selection is of economic and environmental significance. An increasingly accepted measure of feed efficiency is residual feed intake (RFI). Currently, the molecular mechanisms underlying RFI are largely unknown. Additionally, to incorporate RFI into animal breeding programs, feed intake must be recorded on individual pigs, which is costly and time-consuming. Thus, convenient and predictive biomarkers for RFI that can be measured at an early age are greatly desired. In this study, we aimed to explore whether differences exist in the global gene expression profiles of peripheral blood of 35 to 42 day-old pigs with extremely low (more efficient) and high RFI (less efficient) values from two lines that were divergently selected for RFI during the grow-finish phase, to use such information to explore the potential molecular basis of RFI differences, and to initiate development of predictive biomarkers for RFI. RESULTS: We identified 1972 differentially expressed genes (DEGs) (q ≤ 0.15) between the low (n = 15) and high (n = 16) RFI groups of animals by using RNA sequencing technology. We validated 24 of 37 selected DEGs by reverse transcription-quantitative PCR (RT-qPCR) in a joint analysis of 24 (12 per line) of the 31 samples already used for RNA-seq plus 24 (12 per line) novel samples from the same contemporary group of pigs. Using an analysis of the 24 novel samples alone, only nine of the 37 selected DEGs were validated. Genes involved in small molecule biosynthetic process, antigen processing and presentation of peptide antigen via major histocompatibility complex (MHC) class I, and steroid biosynthetic process were overrepresented among DEGs that had higher expression in the low versus high RFI animals. Genes known to function in the proteasome complex or mitochondrion were also significantly enriched among genes with higher expression in the low versus high RFI animals. Alternatively, genes involved in signal transduction, bone mineralization and regulation of phosphorylation were overrepresented among DEGs with lower expression in the low versus high RFI animals. The DEGs significantly overlapped with genes associated with disease, including hyperphagia, eating disorders and mitochondrial diseases (q < 1E-05). A weighted gene co-expression network analysis (WGCNA) identified four co-expression modules that were differentially expressed between the low and high RFI groups. Genes involved in lipid metabolism, regulation of bone mineralization, cellular immunity and response to stimulus were overrepresented within the two modules that were most significantly differentially expressed between the low and high RFI groups. We also found five of the DEGs and one of the co-expression modules were significantly associated with the RFI phenotype of individual animals (q < 0.05). CONCLUSIONS: The post-weaning blood transcriptome was clearly different between the low and high RFI groups. The identified DEGs suggested potential differences in mitochondrial and proteasomal activities, small molecule biosynthetic process, and signal transduction between the two RFI groups and provided potential new insights into the molecular basis of RFI in pigs, although the observed relationship between the post-weaning blood gene expression and RFI phenotype measured during the grow-finish phase was not strong. DEGs and representative genes in co-expression modules that were associated with RFI phenotype provide a preliminary list for developing predictive biomarkers for RFI in pigs.

Detecting Differentially Expressed Genes with RNA-seq Data Using Backward Selection to Account for the Effects of Relevant Covariates.

A common challenge in analysis of transcriptomic data is to identify differentially expressed genes, i.e., genes whose mean transcript abundance levels differ across the levels of a factor of scientific interest. Transcript abundance levels can be measured simultaneously for thousands of genes in multiple biological samples using RNA sequencing (RNA-seq) technology. Part of the variation in RNA-seq measures of transcript abundance may be associated with variation in continuous and/or categorical covariates measured for each experimental unit or RNA sample. Ignoring relevant covariates or modeling the effects of irrelevant covariates can be detrimental to identifying differentially expressed genes. We propose a backward selection strategy for selecting a set of covariates whose effects are accounted for when searching for differentially expressed genes. We illustrate our approach through the analysis of an RNA-seq study intended to identify genes differentially expressed between two lines of pigs divergently selected for residual feed intake. We use simulation to show the advantages of our backward selection procedure over alternative strategies that either ignore or adjust for all measured covariates.