High-resolution genetic mapping of maize pan-genome sequence anchors

In addition to single-nucleotide polymorphisms, structural variation is abundant in many plant genomes. The structural variation across a species can be represented by a 'pan-genome', which is essential to fully understand the genetic control of phenotypes. However, the pan-genome's complexity hinders its accurate assembly via sequence alignment. Here we demonstrate an approach to facilitate pan-genome construction in maize. By performing 18 trillion association tests we map 26 million tags generated by reduced representation sequencing of 14,129 maize inbred lines. Using machine-learning models we select 4.4 million accurately mapped tags as sequence anchors, 1.1 million of which are presence/absence variations. Structural variations exhibit enriched association with phenotypic traits, indicating that it is a significant source of adaptive variation in maize. The ability to efficiently map ultrahigh-density pan-genome sequence anchors enables fine characterization of structural variation and will advance both genetic research and breeding in many crops.