RNA-seq with and without rRNA depletion and varying library prep protocol parameters

Thomas G Brooks 1 Nicholas F Lahens 1 Antonijo Mrčela 1 Shaon Sengupta 1, 2 Peter S Choi 3 Gregory R Grant 1, 4 1. Institute for Translational Medicine and Therapeutics, University of Pennsylvania 2. Division of Neonatology, Children's Hospital of Philadelphia 3. Division of Cancer Pathobiology, Children's Hospital of Philadelphia 4. Department of Genetics, University of Pennsylvania

RNA-seq data remains incompletely understood, despite widespread use. Generation of RNA-seq data involves one of several possible library preparation protocols, each leaving distinct "fingerprints" in the resulting data. In order to asses the affects of individual library preparation steps on the resulting data, we sequenced one sample under different protocols, varying several key factors: fragment size, PCR temperature ramp rate, and mRNA-enrichment method. We also ran a novel "No Select" protocol which avoids the mRNA enrichment step (typically either poly-A selection or Ribo-Zero) by including all RNA and performing extremely deep sequencing to still quantify mRNA abundance. Results identify effects of these protocol parameters on the resulting quantifications and reveal while this No Select method is viable, standard Ribo-Zero protocols capture similar expression levels at lower sequencing depth. The No Select method reveled that patterns of read depth drop-out arise from steps other than the mRNA enrichment, uniquely captured snoRNA expression, and showed that sequenced rRNA content was lower than anticipated. Comparing No Select to poly-A selection and Ribo-Zero in a typical knockout-versus-wildtype mouse experiment reveals that all methods agree on top mRNA differential expression and poly-A selection remains the most efficient way to assess differential expression in mature mRNA abundance while Ribo-Zero or No Select are suitable for measuring pre-mRNA and other RNA abundance.