Identifying barley pan-genome sequence anchors using genetic mapping and machine learning

Shang Gao, Jinran Wu, Jiri Stiller, Zhi Zheng, Meixue Zhou, You-Gan Wang & Chunji Liu

Theoretical and Applied Genetics September 2020; vol. 133: 2535–2544

Key message

We identified 1.844 million barley pan-genome sequence anchors from 12,306 genotypes using genetic mapping and machine learning.

Abstract

There is increasing evidence that genes from a given crop genotype are far to cover all genes in that species; thus, building more comprehensive pan-genomes is of great importance in genetic research and breeding. Obtaining a thousand-genotype scale pan-genome using deep-sequencing data is currently impractical for species like barley which has a huge and highly repetitive genome. To this end, we attempted to identify barley pan-genome sequence anchors from a large quantity of genotype-by-sequencing (GBS) datasets by combining genetic mapping and machine learning algorithms. Based on the GBS sequences from 11,166 domesticated and 1140 wild barley genotypes, we identified 1.844 million pan-genome sequence anchors. Of them, 532,253 were identified as presence/absence variation (PAV) tags. Through aligning these PAV tags to the genome of hulless barley genotype Zangqing320, our analysis resulted in a validation of 83.6% of them from the domesticated genotypes and 88.6% from the wild barley genotypes. Association analyses against flowering time, plant height and kernel size showed that the relative importance of the PAV and non-PAV tags varied for different traits. The pan-genome sequence anchors based on GBS tags can facilitate the construction of a comprehensive pan-genome and greatly assist various genetic studies including identification of structural variation, genetic mapping and breeding in barley.

See https://link.springer.com/article/10.1007/s00122-020-03615-y

Generating barley pan-genome sequence anchors using a strategy combining linkage disequilibrium mapping and machine learning.