Chromosome-scale assembly of European flax (Linum usitatissimum L.) genotypes and pangenomic analysis provide genomic tools to improve breeding

Decoding the genome of cultivated species is one of the key starting points for supporting marker-assisted selection to accelerate breeding programmes. To date, five to six flax genotypes, including fiber flax, linseed and related wild flax, mainly from Canada, China and Russia, have been sequenced and assembled. But no genome assembly was available for Western European flax cultivar, although France is the world's leading producer of fiber flax (85% of world production). The Canadian oilseed flax cultivars CDC Bethune have been mainly used as reference for studies and breeding activities. The best assembly to date was the Chinese fiber Yiya5 assembly, which had the best metric of all the flax genome assemblies available. Recent analyses have shown that up to 30% of fiber flax reads from different origins do not map to the oilseed genome. Thus, much genetic information could be neglected using the oilseed genome as a reference. The aim of this study is to sequence, assemble and annotate the genomes of four European flax cultivars (two fiber Bolchoï and Idéo; two oilseed Marquise and Attila) and provide flax pangenome for genotype characterization and breeding. From 21 to 46 Gb of ONT Minion long reads sequence data were -sequenced for the four genotypes, - assembled using Flye, -polished, validated and organised into hybrid scaffolds using 100 to 197 Gb optical map data (Bionano Genomics), and finally anchored into fifteen T2T pseudomolecules for each genotype using the fiber flax Yiya5 assembly. The final genome size reached 437, 441, 442 and 453 Mb for Idéo, Marquise, Attila and Bolchoï respectively. The four assemblies were annotated using ab initio predictions of genomic sequences and evidence from transcriptomic data. A pangenome graph were constructed for ten cultivars, including the four previous assemblies and six publicly available assemblies (CDC Bethune, Longya10 and Line 3896, Heiya-14, Yiya5 and Atlant). The total pangenome graph size was 835.77 Mbp (74,123 non-redundant orthologous proteins) with the core pangenome size of 172.2 Mbp. The resources generated could be used for the development of genomic tools useful for breeders to improve the efficiency of breeding and variety evaluation.