A case study in RNA-seq reproducibility of Pseudomonas aeruginosa in laboratory models of cystic fibrosis

Across the biological sciences, high throughput gene expression data is used to infer how a organism's gene expression and physiology changes in different environments, cell types, or tissues. While next-generation sequencing technologies such as RNA sequencing (RNA-seq) have enabled scientists to measure genome-wide expression of any organism, but they also require numerous steps where variation could be introduced, making comparison of different data sets from different sources problematic. In this study, we examine reproducibility of RNA-seq data from different laboratories and sequenced with different sequencing pipelines by leveraging RNA-seq datasets generated for the opportunistic human pathogen Pseudomonas aeruginosa in clinical samples and laboratory models of cystic fibrosis. While we found variation in gene expression profiles in samples that were prepared in different laboratories, it was relatively small. The largest source of variation we observed in gene expression profiles resulted from RNA sequencing pipelines. This variation causes different sequencing pipelines to detect different sets of genes that are expressed differently between a synthetic cystic fibrosis sputum medium (SCFM2) and an airway epithelial cell model combined with SCFM2 (epiSCFM2). However, different sequencing pipelines detect many of the same genes with the largest differences in expression between the two models. This study demonstrates that RNA-seq datasets from different sources can be variable and understanding the potential variation between datasets will help investigators to contextualize data they use to strengthen their inferences of biological significance. We also offer recommendations for proper validation of gene expression for downstream mechanistic studies as well as recommendations for standard reporting guidelines for RNA-seq studies.