Background: The de novo assembly of transcriptomes from short shotgun sequences raises challenges due to random and non-random sequencing biases and inherent transcript complexity. We sought to define a pipeline for de novo transcriptome assembly to aid researchers working with emerging model systems where well annotated genome assemblies are not available as a reference. To detail this experimental and computational method, we used early embryos of the sea anemone, Nematostella vectensis, an emerging model system for studies of animal body plan evolution. We performed RNA-seq on embryos up to 24 h of development using Illumina HiSeq technology and evaluated independent de novo assembly methods. The resulting reads were assembled using either the Trinity assembler on all quality controlled reads or both the Velvet and Oases assemblers on reads passing a stringent digital normalization filter. A control set of mRNA standards from the National Institute of Standards and Technology (NIST) was included in our experimental pipeline to invest our transcriptome with quantitative information on absolute transcript levels and to provide additional quality control.
Results: We generated >200 million paired-end reads from directional cDNA libraries representing well over 20 Gb of sequence. The Trinity assembler pipeline, including preliminary quality control steps, resulted in more than 86% of reads aligning with the reference transcriptome thus generated. Nevertheless, digital normalization combined with assembly by Velvet and Oases required far less computing power and decreased processing time while still mapping 82% of reads. We have made the raw sequencing reads and assembled transcriptome publically available.
Conclusions: Nematostella vectensis was chosen for its strategic position in the tree of life for studies into the origins of the animal body plan, however, the challenge of reference-free transcriptome assembly is relevant to all systems for which well annotated gene models and independently verified genome assembly may not be available. To navigate this new territory, we have constructed a pipeline for library preparation and computational analysis for de novo transcriptome assembly. The gene models defined by this reference transcriptome define the set of genes transcribed in early Nematostella development and will provide a valuable dataset for further gene regulatory network investigations.
Keywords: Body plan evolution; Gene regulatory network; Illumina HiSeq; Nematostella embryonic development; Next-generation sequencing; Oases; RNA-seq; Transcriptome; Trinity.