Computational and Statistical Methods for Population Genomics

Genetic diversity in crops represents an essential resource to meet emerging challenges to food security. Here, we describe one such resource; Nested Association Mapping (NAM) populations of rice featuring 10 tropical japonica diversity donors crossed to a common reference IR64 indica.  From each biparental cross, a NAM population of 200 lines was generated by single seed descent selfing. In total, approximately 2000 lines were derived from 10 biparental crosses.

To characterize genetic diversity, 2000 NAM lines were subjected to a modified Genotype by Sequencing (GBS) method.  Variants were called using a standard computational pipeline.  Multiplex sequencing of a large number of individual samples often results in uneven read coverage and missing markers. Therefore, we developed an imputation algorithm, LB-Impute, incorporating a hidden Markov model with emission probabilities adjusted for read coverage.

Next, the final variant dataset was used to examine genetic properties of the NAMs. Segregation distortion is an important issue that can affect the reliability of trait mapping. Analysis of recombination rates predicts the resolution of trait mapping studies.  Our results show that both known and previously unknown segregation distortion loci were identified.

Our ultimate goal is to map tropical agronomic traits. We conducted phenotypic studies for over 20 traits in all 10 NAM populations. The fine mapping resolution and the shuffling of diversity donor alleles across NAMs facilitates within-population Quantitative Trait Loci (QTL) analyses and across-population association analyses. Mapped traits, in addition to germplasm and genotypic data, will be made available to improve rice agriculture in tropical regions.