Optimizing and benchmarking de novo transcriptome sequencing: from library preparation to assembly evaluation

Abstract

BackgroundRNA-seq enables gene expression profiling in selected spatiotemporal windows and yields massive sequence information with relatively low cost and time investment, even for non-model species. However, there remains a large room for optimizing its workflow, in order to take full advantage of continuously developing sequencing capacity.MethodTranscriptome sequencing for three embryonic stages of Madagascar ground gecko (Paroedura picta) was performed with the Illumina platform. The output reads were assembled de novo for reconstructing transcript sequences. In order to evaluate the completeness of transcriptome assemblies, we prepared a reference gene set consisting of vertebrate one-to-one orthologs.ResultTo take advantage of increased read length of >150 nt, we demonstrated shortened RNA fragmentation time, which resulted in a dramatic shift of insert size distribution. To evaluate products of multiple de novo assembly runs incorporating reads with different RNA sources, read lengths, and insert sizes, we introduce a new reference gene set, core vertebrate genes (CVG), consisting of 233 genes that are shared as one-to-one orthologs by all vertebrate genomes examined (29 species)., The completeness assessment performed by the computational pipelines CEGMA and BUSCO referring to CVG, demonstrated higher accuracy and resolution than with the gene set previously established for this purpose. As a result of the assessment with CVG, we have derived the most comprehensive transcript sequence set of the Madagascar ground gecko by means of assembling individual libraries followed by clustering the assembled sequences based on their overall similarities.ConclusionOur results provide several insights into optimizing de novo RNA-seq workflow, including the coordination between library insert size and read length, which manifested in improved connectivity of assemblies. The approach and assembly assessment with CVG demonstrated here would be applicable to transcriptome analysis of other species as well as whole genome analyses.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-015-2007-1) contains supplementary material, which is available to authorized users.

The transcriptome of the veiled chameleon (Chamaeleo calyptratus): A resource for studying the evolution and development of vertebrates

“…When comparing this latter score with other de novo squamate transcriptomes analyzed by means of gVolante, it is only exceeded by one the Madagascar ground gecko (Paroedura picta), which contained 100% of CVG dataset. 22 Notably, our assembly significantly outperforms the previously published Chamaeleo chamaeleon transcriptome, 13 which achieved a modest score of 42.92%. In addition to its completeness, we internally validated our final assembly by mapping reads back to our transcriptome using bwa [v0.7.17] 23 and calculated mapping statistics using bamtools [v2.5.1].…”

“…Furthermore, when compared against the CVG database (233 genes), our assembly possesses 99.14% complete copies of this gene set (ie, missing 2 genes). When comparing this latter score with other de novo squamate transcriptomes analyzed by means of gVolante, it is only exceeded by one the Madagascar ground gecko ( Paroedura picta ), which contained 100% of CVG dataset . Notably, our assembly significantly outperforms the previously published Chamaeleo chamaeleon transcriptome, which achieved a modest score of 42.92%.…”

The whale shark genome reveals patterns of vertebrate gene family evolution

“…Although the BUSCO score of Callorhinchus is higher than in whale shark (which is expected as Callorhinchus was used in selecting the conserved ortholog set for BUSCO), when considering gene families across vertebrates rather than simply the set in BUSCO (See Ancestral Vertebrate Genome Evolution below and Supplementary Note 2), we infer more gene families are missing from Callorhinchus than the whale shark, suggesting that either the whale shark has a greater conservation of the gene families from the ancestral cartilaginous fish than Callorhinchus , or that the whale shark has fewer missing gene families in its assembly than Callorhinchus . When evaluating gene completeness using a rigid 1-to-1 ortholog core vertebrate gene set 26 (implemented in gVolante 27 ), we found 85% core vertebrate genes were complete and found 97.4% core vertebrate genes included partial genes, which compares favorably to other sharks 11 . Thus, the gene content of the whale shark is quite complete for a cartilaginous fish genome and will be informative for questions regarding vertebrate gene evolution.…”

Embryonic development of a parthenogenetic vertebrate, the mourning gecko (Lepidodactylus lugubris)

“…Currently, normal stages of development are characterized for six gecko species, with varying morphologies and ecologies. 6,53,61,62,72 Access to additional resourcesprotocols for husbandry and embryo collection, 18,23 genomic and transcriptomic resources, 19,21,24,103,104 and now manipulatable soft-tissue μCT data across developmentsets the foundation for geckos, and specifically L. lugubris, to be powerful evo-devo models.…”