Genome-wide association study of rice (Oryza sativa L.) inflorescence architecture

Rice yield strongly depends on panicle size and architecture but the genetics underlying these traits and their coordination with environmental cues through various signaling pathways have remained elusive. A genome-wide association study (GWAS) was performed to pinpoint the underlying genetic determinants for rice panicle architecture by analyzing 20 panicle-related traits using a data set consisting of 44,100 SNPs. We defined QTL windows around significant SNPs by the rate of LD decay for each chromosome and used these windows to identify putative candidate genes associated with the trait. Using a publicly available RNA-seq data set we performed analyses to identify the differentially expressed genes between stem and panicle with putative functions in panicle architecture. In total, 52 significant SNPs were identified, corresponding to 41 unique QTLs across the 12 rice chromosomes, with the most signals appearing on chromosome 1 (nine associated SNPs), and seven significant SNPs for each of chromosomes 8 and 12. Some novel genes such as Ankyrin, Duf, Kinesin and Brassinosteroid insensitive were found to be associated with panicle size. A haplotype analysis showed that genetic variation in haplotypes qMIL2 and qNSBBH21 were related to two traits, MIL, the greatest distance between two nodes on the rachis, and NSBBH, the number of primary branches in the bottom half of a panicle, respectively. Analysis of epistatic interactions revealed a marker affecting clustered traits. Several QTLs were identified on different chromosomes for the first time which may explain the phenotypic diversity of rice panicle architecture we observe in our collection of accessions. The identified candidate genes and haplotypes could be used in marker-assisted selection to improve rice yield through gene pyramiding.