Computational protocol: The Identification of Two Head Smut Resistance-Related QTL in Maize by the Joint Approach of Linkage Mapping and Association Analysis

Similar protocols

Protocol publication

[…] For the RIL population, single nucleotide polymorphism (SNP) sites from GBS with minor allele frequency (MAF) <0.05 were first filtered out. Then, the draft parental genotypes were inferred from the low coverage SNP datasets of the RIL population using a maximum parsimonious inference of recombination (MPR) method, and the genotype assignment of each RIL was performed using a hidden Markov model (HMM) approach []. For each RIL, consecutive SNP sites with the same genotype were lumped into blocks, a breakpoint was assumed at the transition between two blocks, and markers co-segregating within a block were combined into a recombination bin []. The genetic map of the RIL population was constructed from bins serving as genetic markers using the R/qtl package function with Haldane map method []. A total of 1638 bins were identified based on the GBS data. Using bin genotypes, the map of the RIL population covers all 10 maize chromosomes with a total genetic distance of 1729.1 cM, and the average bin interval is 1.1 cM. The detailed genetic map used in this study is described in Li et al. (2015) [].QTL analysis for the RIL population and the BC4F1 was conducted with the method of inclusive composite interval mapping (ICIM) in QTL IciMapping software Version 3.3 []. For the BC4F1 population, there were only two genotypes for individuals (HZS/HZS and HZS/QI319), which was similar to the RIL population, and the same method was used for QTL mapping with the BC4F1 population. The LOD threshold of 2.5 was obtained by 1,000 permutations at a significance level of P = 0.05. The correlation of the phenotypic values with different genotypes of the BC5F2 single plants and the BC5F2 families was determined by the module of PROC GLM in SAS []. [...] The best linear unbiased predictions (BLUPs) of head smut disease incidence for the association panel across two environments were calculated with the MIXED procedure in SAS (SAS Institute Inc.), within which genotype, environment and replication were treated as random variables [].For GWAS with 41,819 SNPs, four models, including the general linear model (GLM) with or without PCA and the mixed linear models (MLM) of both K and PCA+K model, were selected to correct for false positives. Both the GLM and MLM models were processed in TASSEL V4.2.1 []. Quantile-quantile plots were shown with a negative log P-value of the observed P-value from the genotype-phenotype association and the expected P-value. The Bonferroni test (0.05/numbers of tests) criterion would be a strict threshold when a large number of markers were used in GWAS []. Thus, a lower threshold of–log10 (P-value) = 5.5 was used as a threshold (P-value < 3.61×10−6). The epistatic interaction between each pair of resistance-associated SNPs was tested by fitting each marker pair to a general linear model as y = mi + mj + mi*mj + error, where y is the head smut disease incidence, mi is the effect of the ith marker within the tested pair, mj is the effect of jth marker, mi*mj is the epistatic effect between the ith and jth marker, and the error is the residual. If there was significant contribution for mi*mj in the model (P<0.05), a significant epistatic marker pair was supposed to be tested. Linkage disequilibrium (LD) within the QTL region was evaluated using the software of Haploview []. […]

Pipeline specifications

Software tools R/qtl, TASSEL, Haploview
Applications WGS analysis, GWAS
Organisms Zea mays