Raw paired-end reads were quality-checked using FastQC (v0.11.9; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) before transcriptomic assembly. Two assemblies were then made, first via the de novo assembly option in CLC Genomics WorkBench v.6.0.4 (CLCBio, Boston, MA) using raw reads, with scaffolding enabled, a minimum contig length of 200 bp, and automatic word and bubble sizes of 24 and 50, respectively. The second was assembled using the Oyster River Protocol (ORP, v2.2.6; [42 (link)]) and default settings (TPM_FILT = 1, STRAND = RF, MEM = 150, CPU = 24), on an Amazon Web Service EC2 server with 32 vCPUs and 128 GB of RAM. The first step in the ORP pipeline is read error correction via Rcorrector (v1.0.3), followed by Illumina adaptor removal and trimming of reads with Phred quality below 3, using Trimmomatic (v0.38). Trinity (v2.8.4; [43 (link)]), Spades55, Spades75 (v3.13.0; [44 (link)]) and Transabyss (v2.0.1; [45 (link)]) are then used to make four de novo assemblies, which are merged via a modified version of Orthofinder (v2.2.6; [46 (link)]) packaged in OrthoFuser [45 (link)]. The CLC and ORP-merged assemblies were assessed via BUSCO (v4.0.6; [47 ]) with the insecta_odb10 database, and TransRate (v1.0.3; [48 (link)]) read mapping.
Free full text: Click here