The Population Genomic Structure of the Asian Cultivated Rice (Oryza sativa L.) I: Genome Sizes, Pan Genomes, and Structural Variation

As the staple food for half the world’s population, rice (Oryza sativa L.) is also well known for its subspecific differentiation and rich within species diversity. As a major international effort to understand the population genomic structure of within O. sativa diversity, a core collection of 3,010 rice accessions from 89 countries were re-sequenced with an average sequencing depth of 14.9x. Based on SNP variation, the 3,010 rice accessions were classified into two major subspecies, 1,764 Xian (Indica) accessions, 800 Geng (Japonica) ones, 221 AUS accessions, 101 aromatic accessions, plus 124 admixtures. Deep analyses of this huge dataset revealed, for the first time, the three important aspects of the population genomic structure of the within O. sativa diversity: genome size (GS) variation, pan genomes (PG) and structural variation (SV). First, de novo assembly of the 3,010 rice genomes produced an average assembled GS of 288.2±21.5Mba and an estimated average GS of 375.2±21.3 Mbp, with a mean repetitive sequence (RS) content of 35.6±3.7%, which is highly correlated with the estimated GS. Secondly, in our PG analyses, a total of 17,893 protein-coding genes were predicted from the 320 Mbp novel sequences of 453 genomes of high quality. These novel genes plus those in the reference Nipponbare genome were merged into 24,070 gene families, of which 11,027 gene families were present in all rice accessions, forming the core PG of O. sativa, while the remaining 13,043 gene families comprise the distributed PG of rice. Using computer simulation, we were able to determine 1,127 undiscovered gene families in the distributed PG of O. sativa. Thus, the PG of O. sativa consists of 25,197 gene families, including 11,027 core gene families (43.8%) and some 14,170 (56.2%) distributed gene families. Classification of the 453 rice accessions based on the presence/absence of the PG gene families was very similar to that based on SNV with very few misclassifications. Thirdly, by directly comparing the contigs of each assembled genome with the Nipponbare and 9311 reference genomes, we discovered large numbers of large SVs. Of these SVs, translocations and deletions were predominant and accounted for most of the SVs detected, while duplications and inversions accounted for small portions of the detected SVs. The detected large SVs were not randomly distributed across the rice genome with many hotspots of SVs identified. Together, our results provided the most comprehensive picture on the population genomic structure of O. sativa and insights into the evolution and origin of this important species.