Workflow of the de novo assembly and resulting annotation steps. (a) Raw reads were filtered and assembled for each sequence pool. (b) Optimized assembly strategies for individual Illumina, 454 and Sanger sequence pools illustrated as number of contigs versus length distribution. Velvet and Oases were used for the Illumina reads, MIRA for the 454 reads and CAP3 and MIRA for Sanger reads. Contigs from individual assemblies were merged by TGICL and CAP3, yielding 120,922 unique transcripts (c) Venn diagram of all assembled transcripts, categorized into transcripts larger 400 bp, transcripts with annotation, and transcripts with peptide identification. GA, genome analyzer; PE, paired end.