Description |
RNA sequencing depicts the comprehensive landscape of the transcriptome. As the methodologies, sequencing technologies, and algorithms continually evolve, ongoing quality control and improvements in RNA-seq are necessary. However, over the past 17 years, quality assessment and benchmarking of transcriptome profiling have predominantly relied on the MAQC resources, characterized by large biological differences. Hence, the understanding of the accuracy and reproducibility of real-world RNA-seq in identifying differential expressions as subtle as those between disease and normal conditions remains lacking. Moreover, the sources of inter-laboratory variations under diverse experimental protocols and bioinformatics pipelines have still not been clarified. To address these, as part of the Quartet project, we tested Quartet samples spiked with ERCC controls and MAQC samples in 45 independent laboratories, each employing its distinct RNA-seq workflow. These RNA-seq data revealed noteworthy variations in multiple aspects of the transcriptome when detecting subtle differential expression in real-world laboratory settings, especially in data quality, absolute gene expression, and differential expression calls. RNA-seq data from 13 laboratories were considered high-quality, covering several distinct experimental protocols, and used for benchmarking analysis pipelines. |