Analytical workflow of double-digest restriction site-associated DNA sequencing based on empirical and in silico optimization in tomato

Double-digest restriction site-associated DNA sequencing (ddRAD-Seq) enables high-throughput genome-wide genotyping with next-generation sequencing technology. Consequently, this method has become popular in plant genetics and breeding. Although computational in silico prediction of restriction sites from the genome sequence is recognized as an effective approach for choosing the restriction enzymes to be used, few reports have evaluated the in silico predictions in actual experimental data. In this study, we designed and demonstrated a workflow for in silico and empirical ddRAD-Seq analysis in tomato, as follows: (i) in silico prediction of optimum restriction enzymes from the reference genome, (ii) verification of the prediction by actual ddRAD-Seq data of four restriction enzyme combinations, (iii) establishment of a computational data processing pipeline for high-confidence single nucleotide polymorphism (SNP) calling, and (iv) validation of SNP accuracy by construction of genetic linkage maps. The quality of SNPs based on de novo assembly reference of the ddRAD-Seq reads was comparable with that of SNPs obtained using the published reference genome of tomato. Comparisons of SNP calls in diverse tomato lines revealed that SNP density in the genome influenced the detectability of SNPs by ddRAD-Seq. In silico prediction prior to actual analysis contributed to optimization of the experimental conditions for ddRAD-Seq, e.g. choices of enzymes and plant materials. Following optimization, this ddRAD-Seq pipeline could help accelerate genetics, genomics, and molecular breeding in both model and non-model plants, including crops.