Abstract
The growing global human population requires improvement in crop production to meet food demand. Crop improvement via breeding can sustainably increase yield production and stability and decrease dependence on fertilisers and pesticides. Recent progresses in pangenomics and machine learning provide opportunities for crop improvement. The development of long-read sequencing technologies is helping overcome challenges in crop genome assembly caused by highly repeated regions or heterozygous sequences. As a result, high-quality crop reference genomes and pangenomes are becoming increasingly accessible, enhancing downstream analyses such as variant discovery and association mapping, which are crucial for identifying breeding targets for crop improvement. Machine learning approaches help to characterise the growing volume of plant genomic data and facilitating real-time high-throughput phenotyping of agronomic traits. Moreover, crop databases that integrate the increasing amount of genotypes identified using pangenomes and machine learning approaches are valuable for uncovering novel trait-associated candidate genes. With an increasing understanding of crop genetics, genomic selection and genome editing emerge as powerful tools for cultivating crops that are resistant to both biotic and abiotic stresses, while also achieving high productivity.
