The advent of next-generation DNA sequencing (NGS) technologies has led to the development of rapid genome-wide Single Nucleotide Polymorphism (SNP) detection applications in various plant species. Recent improvements in sequencing throughput combined with an overall decrease in costs per gigabase of sequence is allowing NGS to be applied to not only the evaluation of small subsets of parental inbred lines, but also the mapping and characterization of traits of interest in much larger populations. Such an approach, where sequences are used simultaneously to detect and score SNPs, therefore bypassing the entire marker assay development stage, is known as genotyping-by-sequencing (GBS). This review will summarize the current state of GBS in plants and the promises it holds as a genome-wide genotyping application.

The use of next-generation DNA sequencing technologies has greatly facilitated reference-guided variant detection in complex plant genomes. However, complications may arise when regions adjacent to a read of interest are used for marker assay development, or when reference sequences are incomplete, as short reads alone may not be long enough to ascertain their uniqueness. Here, the possibility of generating longer sequences in discrete regions of the large and complex genome of maize is demonstrated, using a modified version of a paired-end RAD library construction strategy. Reads are generated from DNA fragments first digested with a methylation-sensitive restriction endonuclease, sheared, enriched with biotin and a selective PCR amplification step, and then sequenced at both ends. Sequences are locally assembled into contigs by subgrouping pairs based on the identity of the read anchored by the restriction site. This strategy applied to two maize inbred lines (B14 and B73) generated 183,609 and 129,018 contigs, respectively, out of which at least 76% were >200 bps in length. A subset of putative single nucleotide polymorphisms from contigs aligning to the B73 reference genome with at least one mismatch was resequenced, and 90% of those in B14 were confirmed, indicating that this method is a potent approach for variant detection and marker development in species with complex genomes or lacking extensive reference sequences.