PT  - JOURNAL ARTICLE
AU  - Yuguang Xiong
AU  - Magali Soumillon
AU  - Jie Wu
AU  - Jens Hansen
AU  - Bin Hu
AU  - Johan G.C. van Hasselt
AU  - Gomathi Jayaraman
AU  - Ryan Lim
AU  - Mehdi Bouhaddou
AU  - Loren Ornelas
AU  - Jim Bochicchio
AU  - Lindsay Lenaeus
AU  - Jennifer Stocksdale
AU  - Jaehee Shim
AU  - Emilda Gomez
AU  - Dhruv Sareen
AU  - Clive Svendsen
AU  - Leslie M. Thompson
AU  - Milind Mahajan
AU  - Ravi Iyengar
AU  - Eric A. Sobie
AU  - Evren U. Azeloglu
AU  - Marc R. Birtwistle
TI  - A Comparison of mRNA Sequencing with Random Primed and 3’-Directed Libraries
AID  - 10.1101/098905
DP  - 2017 Jan 01
TA  - bioRxiv
PG  - 098905
4099  - http://biorxiv.org/content/early/2017/01/06/098905.short
4100  - http://biorxiv.org/content/early/2017/01/06/098905.full
AB  - Deep mRNA sequencing (mRNAseq) is the state-of-the-art for whole transcriptome measurements. A key step is creating a library of cDNA sequencing fragments from RNA. This is generally done by random priming, creating multiple sequencing fragments along the length of each transcript. A 3’ end-focused library approach cannot detect differential splicing, but has potentially higher throughput at lower cost (~10-fold lower), along with the ability to improve quantification by using transcript molecule counting with unique molecular identifiers (UMI) to correct for PCR bias. Here, we compare implementation of such a 3’-digital gene expression (3’-DGE) approach with “conventional” random primed mRNAseq, which has not yet been done. We find that while conventional mRNAseq detects ~15% more genes, the resulting lists of differentially expressed genes and therefore biological conclusions and gene signatures are highly concordant between the two techniques. We also find good quantitative agreement on the level of individual genes between the two techniques in terms of both read counts and fold change between two conditions. We conclude that for high-throughput applications, the potential cost savings associated with the 3’-DGE approach are a very reasonable tradeoff for modest reduction in sensitivity and inability to observe alternative splicing, and should enable much larger scale studies focused on not only differential expression analysis, but also quantitative transcriptome profiling. The computational scripts and programs, along with experimental standard operating procedures used in our pipeline presented here, are freely available on our website (www.dtoxs.org).