Fig. 5

Schematic workflow of the steps performed for de novo transcriptome assembly of Oscarella lobularis. (1) Illumina sequencing was performed on biological samples of 40 pooled clonal stage 3 buds exposed to the same conditions: dissociated in CMFSW and reaggregated in NSW. 454 sequencing was performed on tissue from one adult undergoing sexual reproduction containing a mixed population of embryos and pre-larvae collected in the bay of Marseille [124]. Illumina reads were quality controlled using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Low-quality reads and adapters were removed using TrimGalore (https://github.com/FelixKrueger/TrimGalore). Illumina and 454 sequencing libraries were first de novo assembled individually and then merged. lllumina sequences were de novo assembled using Trinity. 454 reads were de novo assembled with Staden and GAP4 [13] by Eurofins. (2) The two assemblies were then concatenated. (3) Redundancy was reduced by clustering contigs > 80% identity with CD-hit EST [87] retaining the longer contig. (4) To remove sequence redundancy further, we indexed and mapped the contigs with GMAP [141] with a 95% sequence identity. To remove contaminants, we used Vecscreen and BlastN. (5) Using Transdecoder (https://github.com/TransDecoder) with the Transdecoder.predict option we were able to identify the most likely protein sequences encoded in open reading frames (ORFs) ≥ 100 aa. For additional experimental details see the Material and Methods section