Rhazya stricta Transcriptome sequencing. Rhazya stricta

Three assemblers were used for transcriptome reconstruction of PRJNA213260, CAP3 (Huang & Madan, 1999), MIRA (Chevreux et al, 2004) and GS De Novo Assembler (Roche Basel, Switzerland) (all default). To merge all three assemblers, CAP3 was used with increased stringency -o 40 and -p 90, to create the 454 super assembly. Illumina assembly: The PRJNA213261 data was divided into twelve pools (per plant and leaf type), then assembled first using Velvet (Zerbino & Birney, 2008) at two K-mers 37 and 43. Then Oases (Schulz et al, 2012), was applied to the output from Velvet (both K-mers). Using the output from Oases, a Perl script was designed to select the 'longest' transcript when multiple isoforms are produced per locus (see Oases; Schulz et al, (2012)). The 'longest' contigs from the 37 and 43 k-mer assemblies were then combined using CAP3 (default settings). To merge the twelve individual assemblies a clustering strategy was used where all twelve assemblies were concatenated into a single file. Once concatenated CD- HIT-EST (Weizhong & Godzik, 2006) was used with the following parameters: aS 0.4 -s 0.7 -aL 0.5. Clusters which had at least three contigs were selected using an in-house Perl script, to create the Illumina super assembly. Hybrid assembly: The 454 super assembly and Illumina super assembly were concatenated then clustered using CD-HIT-EST default settings (90% similarity). The representative cluster contigs were then split back into the assemblies which they were derived from (454 or Illumina), using an in-house Perl script. The Illumina assembly was then used to make a BLAST (Altschul et al, 1990) database, to which the 454 contigs were searched using BLAST to find alignments with a minimum E-value 1e-20. 454 clusters without significant BLAST hits were then extracted and concatenated with the Illumina clusters to produce the final transcriptome.