Comparing control options for time-series RNA sequencing experiments in nonmodel organisms: An example from grasses

Abstract

RNA sequencing (RNA-seq) is a widely used approach to investigate gene expression and increasingly is used in time-course studies to characterize transcriptomic changes over time. Two primary options are available as controls in time-course experiments: samples collected at the first sampling time are used as controls (temporal control, TC) and samples collected in parallel at each individual sampling time are used as controls (biological control, BC). While both approaches are used in experimental studies, we know of no analyses performed to date that directly compare effects of control type choices on identifying differentially expressed genes (DEGs) and subsequent functional analysis. In the current study, we compare experimental results using these different control types for time-course RNA-seq drought stress experiments in two wild grass species in the genus Paspalum. Our results showed BC assemblies gave a higher number of loci in both species. The number of DEGs increased with increasing stress and then decreased dramatically at the recovery time point using both control types. Expression levels of the same DEGs were highly correlated between control types in both species, ranging from r = .653 to r = .852. We also observed similar rank orders of shared enriched Gene Ontology term lists using the two different control types. Collectively, our findings suggest similar results in differential gene expression and functional annotation between control types. The ultimate choice of control type will rely on the experimental length and organism type, with labour time and sequencing costs as additional factors to be considered.

Keywords: RNA-seq; RNA-seq control; differentially expressed gene; functional annotation; time-course experiment.