Sequencing

11th Annual Sequencing, Finishing, and Analysis in the Future Meeting
ROBUST GENOME-WIDE TRANSCRIPTOME
ENRICHMENT SEQUENCING FOR RESEARCH AND
CLINICAL PLATFORMS
Friday, 3rd June 15:00 La Fonda Ballroom Talk (OS‐9.04)
Harsha Doddapaneni 1 , Prof. Jianhong Hu 1 , Hsu Chao 2 , Simon White 2 , Tittu Matthew 2 ,
Viktoriya Korchina 2 , Caitlin Nessner 2 , Sandra Lee 2 , Donald W Parsons 3 , David A Wheeler 2 ,
Angshumoy Roy 4 , Eric Boerwinkle 5 , Donna Muzny 1 , Richard Gibbs 2
1 Human Genome Sequencing Center Baylor College of Medicine, 2 Human Genome Sequencing
Center, Baylor College of Medicine, 3 Department of Pediatrics, Baylor College of Medicine and
Texas Children’s Hospital, 4 Department of Pathology & Immunology, Baylor College of Medicine;
Department of Pediatrics, Texas Children’s Hospital, 5 University of Texas Health Science Center
at Houston
Transcriptome sequencing (RNA‐Seq) together with whole genome sequencing (WGS) offers an in‐ tegrated
informative dataset for functional characterization of human transcriptome. Generation of high‐quality data is
essential for success in RNA‐Seq studies. However, standard RNA‐Seq methods rely heavily on very high quality
RNA samples, and the library preparation involves multiple enzy‐ matic manipulations as RNA is first converted to
double‐stranded cDNA and then into paired‐end libraries. Therefore to increase RNA‐Seq utility in routine clinical
setting, these protocols have to overcome RNA quality constraints as well as need rapid preparation protocols.
Since August 2010, at BCM‐HGSC we have generated transcriptome data for 4519 samples in support of different
cancer, pharmacogenomics, vascular and other disease projects. These RNA‐Seq libraries are strand‐specific and
poly(A)+ enriched and use automated library preparation workflow. To monitor sample and process variability, in
addition to sequence metrics, we use RNA developed by the External RNA Controls Consortium (ERCC). Our primary
RNA‐Seq data analysis pipeline is built on STAR and Cuff‐Links software and we assess performance of individual
RNA‐Seq libraries by 12 different metrics for library process consistency.
Since 2014, we have supported exome‐capture transcriptome‐seq by combining our strand‐specific RNA‐Seq
protocol with whole exome sequencing protocol as tool to handle low quality RNA and RNA extracted from FFPE
specimens. Total RNA (40 ng) from control samples as well as FFPE samples from cancer patients (2‐3 RIN and DV
200 of >30%) sequenced using this protocol have shown that sensitivity of this approach for detecting fusions and
somatic SNVs is comparable to that of standard poly‐A+ enriched libraries.
Our recent efforts are focused on designing ways in which we can simplify the RNA‐Seq protocol to reduce sample
preparation time as well as increase sample throughput. In this regard, we have developed a RNA‐Seq protocol using
first strand cDNA as a template for preparing libraries using Accel‐NGS 1S DNA Library Kit from Swift Biosciences.
Libraries prepared using Universal Hu‐ man Reference (UHR) RNA and human Placenta RNA with this protocol
generated 80‐120 million reads/sample with high unique rates (89 % for UHR and 75 % for Placenta). Comparison
of gene ex‐ pression values as Fragments Per Kilobase of transcript per Million mapped reads (FPKM) between 1S
and RNA‐Seq protocol gave a correlation of 0.97 for UHR sample and 0.97 for Placenta sample. Over all, this
workflow eliminates the need to generate second cDNA synthesis and also reduces the library preparation time by
half over standard poly‐A+ enriched workflow. This protocol also allows to multiplex samples after ligating the
molecular barcodes to the samples, but before PCR to increase sample throughput.
In all, these improvements to our RNA‐Seq workflows will allows us to handle low quality RNA including RNA from
FFPE specimens as well as result in fast turnaround times to have practical utility in both research and clinical
settings.
146