Abstract:

Genome-wide association studies (GWAS) take genome-wide high-density molecular markers to identify associations between genotype and phenotype, which have been widely used for genetic dissection of quantitative traits in plants and animals. However, previous GWAS methods focused on finding a handful of major loci and were not able to detect multi-allelic genetic variation in natural populations based on bi-allelic SNP marker, which caused limitations in extending application of GWAS. The restricted two-stage multi-locus genome-wide association analysis (RTM-GWAS) firstly groups multiple adjacent and tightly linked SNPs based on linkage disequilibrium to form multi-allelic SNPLDB markers with multiple haplotypes as alleles. Secondly, population structure bias is estimated using the genetic similarity coefficient matrix calculated from SNPLDB marker, and the eigenvectors of the similarity matrix are extracted and incorporated as model covariates to correct for population structure bias and to reduce false positives. Finally, RTM-GWAS utilizes two-stage association analysis to detect genome-wide QTLs and their multiple alleles efficiently based on the SNPLDB marker and multi-locus multi-allele model, and builds the final multi-QTL genetic model with the total QTL genetic contribution restricted to trait heritability. RTM-GWAS can also detect QTL-by-environment interaction effect using plot-based phenotype data, and can detect not only the main effect QTL, but also QTL with only interaction effect with environment. RTM-GWAS solves the issue that multiple alleles are not estimable in previous GWAS, and also improves the detection power and reduces the false positive rate by fitting multiple QTLs simultaneously in a multi-locus model. It provides a potential solution for a relatively thorough detection of genome-wide QTLs and their multiple alleles, and the allele effect and relative frequency can also be estimated. From RTM-GWAS results, a QTL-allele matrix can be constructed as a compact form of the population genetic constitution, and can be further used for gene discovery. QTL-allele matrix also provides a new tool for studies on the dynamic change of QTLs and their multiple alleles (genes and their multiple alleles), such as population genetic differentiation and population-specific and new alleles. According to QTL-allele matrix, the progeny genotype of cross between parental lines can be simulated by using computer simulation, and then the phenotype can be predicted to assist optimal cross design and molecular design breeding. In addition, RTM-GWAS is more efficient in QTL detection for bi-parental recombinant inbred line population and multi-parental nested association mapping population because the population structure bias can be well-controlled. The present paper presents the principles and procedures of the RTM-GWAS method at first, and then provides some potential applications of RTM-GWAS in plant genetic and breeding studies.

Key words: restricted two-stage multi-locus genome-wide association analysis, multiple alleles, SNPLDB marker, multi-locus model, QTL-allele matrix