Next-generation sequencing can provide access to the genomic sequence of even large and complex plant genomes. Three major strategies exist to assess the genomic information of a species at different scales and complexity levels: transcriptome, target capture and whole-genome shotgun sequencing. The scope of this thesis was to evaluate each concept, ascertain its potential for the discovery of genetic diversity, and develop methods for their improvement. With these objectives, the economically important crops rye, maize and barley were investigated to reveal novel insights into their genetic diversity. The study constructed the first rye transcriptome reference that was utilized for variant discovery revealing ~18,000 single nucleotide variants (SNVs) in coding regions. Subsequently, this resource was converted into a genotyping assay (RYE5k) for application e.g. in breeding programs. The identification of genomic variants requires a high degree of accuracy. Two methods were developed to increase the accuracy in the process of variant discovery: the ‘combinatorial variant calling’ and the approach of ‘k-mer repeat investigation’. With the first method, the reliability of variant calling was increased by the interlaced support and analysis of multiple detection procedures. The approach was successfully applied to determine the diversity in biomass-related genes of maize. Hereby, the applied capture sequencing approach revealed 86,875 SNVs in coding regions. The second method was motivated by the complexity of the large and repetitive barley genome. Therefore, k-mer analyses were used to gain knowledge of repetitive features and this resulted in greater precision in variant calling. The positive effect was shown in a genome-wide diversity study of barley. As a result, more than 15 million high-quality SNVs were identified in five cultivars and a wild progenitor of cultivated barley. The study successfully revealed novel insights into the genetic diversity of barley.