Dissecting Unused Natural Variation in Rice for Yield Associated Traits Using High Throughput Sequencing

We will have to produce significantly more rice in less time to feed nine billion people without adversely affecting the resource base. For achieving this growth, several approaches including introgression of desirable alleles from the wild species need to be followed. Around 2000 BILs (Backcross Inbred Lines) were developed from six AA genome species in background of indica cultivar and evaluated for desirable traits for three years. Based on evaluation, 350 BILs of three populations derived from O.rufipogon, O.glumaepatula and O.longistaminata were selected for mapping of yield contributing traits. Selected BILs were phenotyped for desirable traits in different environments and ANOVA indicated a significant variation. All BILs with their respective parents were sequenced following Genotype by sequencing on Illumina Hiseq2500. A total of 91.5 million reads were generated and aligned to 93-11 reference genome using BWA. SNPs were called using Samtools and GATK and those common between two were filtered to find potential polymorphic SNPs.  75000 common SNPs were called for parents of O.rufipogon population, of which 15000 were used as potential sites where each BIL was genotyped. Missing genotypes were imputed by k-nearest neighbour algorithm. Genetic distance was calculated by modified sliding window approach. QTL mapping was carried out using QTLIciMapping-v3.1. One major QTL for grain weight found at chromosome 5 explaining 45% of phenotype with negative additive effect. Similar strategy is being applied to find QTL on other two populations. Derived QTL from all populations will be used to dissect consensus QTL for desirable traits in wild species.