Protocol dependence of sequencing-based gene expression measurements

Tal Raz, Philipp Kapranov, Doron Lipson, Stan Letovsky, Patrice M. Milos, John F. Thompson

Research output: Contribution to journalArticlepeer-review

76 Scopus citations

Abstract

RNA Seq provides unparalleled levels of information about the transcriptome including precise expression levels over a wide dynamic range. It is essential to understand how technical variation impacts the quality and interpretability of results, how potential errors could be introduced by the protocol, how the source of RNA affects transcript detection, and how all of these variations can impact the conclusions drawn. Multiple human RNA samples were used to assess RNA fragmentation, RNA fractionation, cDNA synthesis, and single versus multiple tag counting. Though protocols employing polyA RNA selection generate the highest number of non-ribosomal reads and the most precise measurements for coding transcripts, such protocols were found to detect only a fraction of the non-ribosomal RNA in human cells. PolyA RNA excludes thousands of annotated and even more unannotated transcripts, resulting in an incomplete view of the transcriptome. Ribosomal-depleted RNA provides a more cost-effective method for generating complete transcriptome coverage. Expression measurements using single tag counting provided advantages for assessing gene expression and for detecting short RNAs relative to multi-read protocols. Detection of short RNAs was also hampered by RNA fragmentation. Thus, this work will help researchers choose from among a range of options when analyzing gene expression, each with its own advantages and disadvantages.

Original languageAmerican English
Article numbere19287
JournalPLoS ONE
Volume6
Issue number5
DOIs
StatePublished - 2011
Externally publishedYes

Fingerprint

Dive into the research topics of 'Protocol dependence of sequencing-based gene expression measurements'. Together they form a unique fingerprint.

Cite this