<p>There is an urgent need to increase sustainable crop production. The application of molecular marker technologies such as genomic selection and machine learning based approaches are aiding accelerated crop improvement. Conventional molecular marker technologies use single nucleotide polymorphisms to predict traits, however these do not capture local epistasis and can be challenging for machine learning applications. With the growth of genome sequence data, it is possible to define haplotypes that can account for local epistatic effects and are more suitable for machine learning models. This review discusses the different methods for defining haplotype blocks and their application in plant breeding.</p>