10k‐level integrated rice database shows power for exploiting rare variants

Rice is one of the most important staple crops, feeding half of the world's population. In the 1990s, rice became a model plant to study monocots due to its relatively small genome and efficient transformation system (Izawa and Shimamoto, 1996). Rice is the first crop with a sequenced and assembled reference genome (Wang and Han, 2022), and rice genes have been cloned via the map-based cloning system, showing that key regulators in rice play critical roles in determining important agronomic traits (Guo et al., 2018; Chen et al., 2022). Meanwhile, the history of rice domestication could be traced to 10,000 years ago spreading in many countries around the world, providing abundant germplasm resources. With the development of genome sequencing technology, different groups have collected germplasm resources worldwide and performed genome-wide association studies (GWAS) and pan-genome analyses, which revealed the genomic loci associated with the phenotypic differences among these varieties (Huang et al., 2010; Wang et al., 2018; Qin et al., 2021; Shang et al., 2022). Based on these findings in functional genomics, the traits that were expensive or laborious to examine and thus hardly to be modified by traditional breeding for a long time, such as grain quality and stress tolerance, were successfully improved in a short time by tracing the beneficial alleles of cloned genes (Qian et al., 2016; Wei et al., 2021). The breeding strategy of gene replacement could be used to efficiently achieve a specific breeding goal; however, new challenges have emerged as the requirements for breeding have increased to improve comprehensive traits. First, the pleiotropic effects of genes lead to the unavoidable trade-off effect of allele replacement. Second, the beneficial effect of certain alleles usually relies on certain genetic backgrounds and environmental conditions. Third, the sequenced rice genomes were not integrated. These require deeper research in: (i) identifying better alleles of cloned genes, especially for the rare alleles; (ii) a better and more comprehensive understanding of the genetic background of rice varieties; and (iii) constructing an integrated and convenient database to serve both the basic research and breeding. A recent study by Shang Lianguang and Qian Qian's groups collected the resequencing data published in 19 studies, which covered 10,548 accessions of Asian cultivated rice Oryza sativa and wild rice Oryza rufipogon from 98 countries worldwide, and reanalyzed them together to generate a super-large rice genomic variation dataset (Wang et al., 2023). This work landmarks a 10k-level population of rice genomes with an integrated map of genetic variations, together with the newly identified large number of rare mutations and a convenient online database. With this integrated map, the authors found a huge number of variations, including 54,378,986 single nucleotide polymorphisms (SNPs) and 11,119,947 insertions/deletions (InDels), indicating that in average every seven base pairs contain one SNP and every 33 base pairs contain one InDel in these 10k rice populations. Among them, 84% of SNPs and 92% of InDels were rare with a minor allele frequency (MAF) lower than 0.01, which were hard to be identified by a relatively small population (Figure 1). Indeed, by randomly subsampling the population, the authors found that the numbers of SNPs and Indels started to saturate until the sample size reached 10,000. Of these rare mutations, 12.65% were non-synonymous mutations located in genes and may be functional. In addition, the authors also revealed 254,051 presence/absence variations (PAVs) from an integrated graph-based pan-genome generated from 356 rice accessions, and 184,736 PAVs from a 10k population using this pan-genome as a reference. The integration and reanalysis of these rice population sequencing data empowered the better and saturated understanding of the genetic diversity in the rice genome. More importantly, the authors established an online database called RSPVM (rice super-population variation map, at http://www.ricesuperpir.com/web/rspvm). The database contained six sections, including the information of all samples, a query and view service for SNPs and InDels, a query and view service for PAVs, the variation frequency for different rice populations, a download service for rare variations, and a series of tools for analyzing variants. The authors showed in certain cases that this super-large dataset together with the information on the rice population could be used in deep analysis for the allelic variations on interested genes. For example, the PAV patterns of rice cadmium resistance gene OsLCD were diverse among different cultivated and wild subpopulations. A rare absence variation of the rice high-affinity nitrate transporter protein coding gene OsNRT2.4 was identified in several accessions. Moreover, using the complete information of variations, 4,522 and 6,923 haplotypes were revealed for grain length and quality-controlling genes GL7/GW7 and GW8, respectively, which were then clustered into five groups for each gene based on a neighbor-joining tree. The rice accessions with different haplotype groups of GL7 and GW8 indeed showed significant differences in grain-related traits. With the online RSPVM database, users can easily conduct similar analyses. Taken together, Wang et al. (2023) provided a comprehensive integrated genomic variation analysis of 10k-level rice populations, providing a much deeper understanding of rice genetic diversity with revealed rare mutations. The authors also built a valuable online database, which will largely benefit scientists studying rice gene functions, and breeders in cultivating new rice varieties. Taking the advances of sequencing technologies and the great efforts of the whole rice research community, the rice genomic study comes to the post-10k-population era and genomes are no longer the bottleneck. Still, it is very challenging to fully utilize these big data and their germplasms, especially to functionally annotate these variations. In the future, it calls for more technical advances and comprehensive studies in phenotyping, envirotying, and specific artificial intelligence algorithms, to bring rice breeding to the next era. The authors declare no conflicts of interest.