Genetic association analysis and meta-analysis of imputed SNPs in longitudinal studies

Genome-wide association study of agronomic traits in bread wheat reveals novel putative alleles for future breeding programs

BACKGROUND

Identification of loci for agronomic traits and characterization of their genetic architecture are crucial in marker-assisted selection (MAS). Genome-wide association studies (GWAS) have increasingly been used as potent tools in identifying marker-trait associations (MTAs). The introduction of new adaptive alleles in the diverse genetic backgrounds may help to improve grain yield of old or newly developed varieties of wheat to balance supply and demand throughout the world. Landraces collected from different climate zones can be an invaluable resource for such adaptive alleles.

RESULTS

GWAS was performed using a collection of 298 Iranian bread wheat varieties and landraces to explore the genetic basis of agronomic traits during 2016-2018 cropping seasons under normal (well-watered) and stressed (rain-fed) conditions. A high-quality genotyping by sequencing (GBS) dataset was obtained using either all original single nucleotide polymorphism (SNP, 10938 SNPs) or with additional imputation (46,862 SNPs) based on W7984 reference genome. The results confirm that the B genome carries the highest number of significant marker pairs in both varieties (49,880, 27.37%) and landraces (55,086, 28.99%). The strongest linkage disequilibrium (LD) between pairs of markers was observed on chromosome 2D (0.296). LD decay was lower in the D genome, compared to the A and B genomes. Association mapping under two tested environments yielded a total of 313 and 394 significant (-log10 P >3) MTAs for the original and imputed SNP data sets, respectively. Gene ontology results showed that 27 and 27.5% of MTAs of SNPs in the original set were located in protein-coding regions for well-watered and rain-fed conditions, respectively. While, for the imputed data set 22.6 and 16.6% of MTAs represented in protein-coding genes for the well-watered and rain-fed conditions, respectively.

CONCLUSIONS

Our finding suggests that Iranian bread wheat landraces harbor valuable alleles that are adaptive under drought stress conditions. MTAs located within coding genes can be utilized in genome-based breeding of new wheat varieties. Although imputation of missing data increased the number of MTAs, the fraction of these MTAs located in coding genes were decreased across the different sub-genomes.

Interaction association analysis of imputed SNPs in case-control and follow-up studies

A new method is described to assess the interactions of imputed SNPs (single nucleotide polymorphisms) in case-control and follow-up studies, properly incorporating SNP imputation uncertainty in the likelihood model. Using simulation studies and analysis of real data obtained from the Framingham study cohort, we compare the performance of this new method to DOSAGE and NAIVE (also known as Best-Guess) methods, developed and commonly used in the context of single SNP and extended to SNP-by-SNP interaction. The results show that only our new method is unbiased under all examined scenarios regarding allele frequencies, imputation uncertainty degree, and interaction effect size. In addition, our method achieves at least as much power as the other two, and exceeds their statistical power in certain follow-up analysis situations. This method is fast enough to perform Genome Wide Interaction Studies (GWIS) with hundreds of thousands of interactions. By performing an exhaustive simulation study let us to provide recommendations for selecting the most appropriated method depending on MAF, interaction effect size, and uncertainty degree. In general, DOSAGE and our proposed method are recommended in most situations being our method more powerful and accurate when uncertainty and effect increase.