Transcriptome profiling of leaves and roots from rooibos (aspalathus linearis) using oxford nanopore sequencing

Abstract

Rooibos (Aspalathus linearis) is one of the few endemic South African plants that has achieved economic importance and international acclaim, mostly as a herbal tea. Plant production, limited to a small mountainous region in South Africa, is at risk as commercial rooibos longevity is in decline, mostly due to low stress tolerance. Transcriptome data can serve to identify molecular markers for improved stress response, which would speed up selection and facilitate the establishment of breeding programmes. Previously, rooibos leaf transcriptomes have been sequenced using Illumina, which yields short reads, hampering correct reassembly of full-length transcripts. Here, we established Oxford Nanopore-based, long-read transcriptome analysis for leaf and root samples from rooibos. We report on potential pitfalls in data pre-processing (PolyA tail trimming and rRNA removal), and compare two assemblers (RATTLE and RNA-Bloom2) and two clustering algorithms (VSEARCH and CD-HIT). The best assembly comprising 169,122 transcripts was generated using RNA-Bloom2 with short-read polishing, followed by CD-HIT clustering. Of the 95,054 predicted proteins, only 67% were also present in the Illumina dataset. The remainder comprised substantially shorter, mostly full-length sequences from a wide range of primary and secondary biosynthesis pathways. Functional annotation indicated that this transcriptome represents a high-quality, comprehensive resource for data mining.