Transcriptome Assembly and Evaluation, using Sequencing Quality Control (SEQC) Data

The US Food and Drug Administration (FDA) has coordinated the Sequencing Quality Control project (SEQC/MAQC-III) with the goal of assessing the technical performance of RNA-Seq experiments comprehensively. The SEQC consortium has generated benchmark datasets of well-studied reference samples, sequenced at multiple sites, and using different sequencing platforms, with controlled settings. The generated RNA-Seq data is used in separate studies to measure quality metrics, spike-in controls, limits of detection, effects of analytic pipeline and assessments of RNA-Seq accuracy and reproducibility. All samples were distributed among six independent centers in the study: 1Australian Genome Research Facility (AGR), 2Beijing Genomics Institute (BGI), 3Weill Cornell Medical College (CNL), 4City of Hope (COH), 5Mayo Clinic (MAY) and 6Novartis (NVS). The SEQC uses the samples from MAQC I consortium (Shi et. al., 2006): sample A is the well-characterized Universal Human Reference RNA (UHRR), and B is Human Brain Reference RNA (HBRR). The synthetic RNA from the External RNA Control Consortium (ERCC) (Baker et. al, 2005) was spiked in. Samples C and D were generated by mixing samples A and B in ratios of 3:1 and 1:3, respectively. Each one of samples A and B had 5 replicates. Replicates 1 to 4 were prepared in each site. The vendor prepared the fifth replicate. To examine the effect of the instrument on the RNA-Seq experiments, all the samples were sequenced using Illumina’s HiSeq 2000, and for generating longer reads three sites sequenced samples A and B using the Roche 454 GS FLX platform. The other next generation sequencing technology, SOLiD, was also used. The SEQC consortium overall sequenced 108 libraries on a HiSeq 2000, 68 libraries on SOLiD, and 6 libraries on a Roche 454, generating more than 100 billion reads, for samples A to D.