Strategies for transcriptome analysis in nonmodel plants

Even with recent reductions in sequencing costs, most plants lack the genomic resources required for successful short-read transcriptome analyses as performed routinely in model species. Several approaches for the analysis of short-read transcriptome data are reviewed for nonmodel species for which the genome of a close relative is used as the reference genome. Two approaches using a data set from Phytophthora-challenged Rubus idaeus (red raspberry) are compared. Over 70000000 86-nt Illumina reads derived from R. idaeus roots were aligned to the Fragaria vesca genome using publicly available informatics tools (Bowtie/TopHat and Cufflinks). Alignment identified 16956 putatively expressed genes. De novo assembly was performed with the same data set and a publicly available transcriptome assembler (Trinity). A BLAST search with a maximum e-value threshold of 1.0 × 10(-3) revealed that over 36000 transcripts had matches to plants and over 500 to Phytophthora. Gene expression estimates from alignment to F. vesca and de novo assembly were compared for raspberry (Pearson's correlation = 0.730). Together, alignment to the genome of a close relative and de novo assembly constitute a powerful method of transcriptome analysis in nonmodel organisms. Alignment to the genome of a close relative provides a framework for differential expression testing if alignments are made to the predefined gene-space of a close relative and de novo assembly provides a more robust method of identifying unique sequences and sequences from other organisms in a system. These methods are considered experimental in nonmodel systems, but can be used to generate resources and specific testable hypotheses.