Skip to main content
. 2016 Dec 14;5:e22502. doi: 10.7554/eLife.22502

Figure 2. GWAS for primary seed dormancy on a set of 161 A. thaliana Swedish accessions.

(A) Manhattan plot of genome-wide association results for GR21. The dotted horizontal line indicates a significance level of 0.05 after Bonferroni correction for multiple testing. (B) Magnification of the peak region on chromosome 5. The most significant SNP (SNP1) was located at position 18,570,773. SNP2 (18,592,365) is in moderate LD with SNP1, despite a physical separation of almost 21 kb. Both SNPs are highlighted. (C) Output of a local multi-locus association scan using SNP2 as starting point. The resulting optimal model consisted of three highlighted SNPs at the DOG1 locus: SNP2, SNP3 (18,591,702) and SNP4 (18,590,289). (D) Annotated genes in the region under the peak. The GWAS results can be viewed interactively online (https://goo.gl/30EPt3).

DOI: http://dx.doi.org/10.7554/eLife.22502.004

Figure 2.

Figure 2—figure supplement 1. The effect of mee62, cipk19 and cipk20 knock-outs on dormancy.

Figure 2—figure supplement 1.

Effects were assessed by measuring the germination rate of insertion mutants in response to various length of after-ripening (0, 7 and 21 days). Col-0 and the dog1-2 mutant were included as controls. At each time point, we analysed three biological replicates per genotype (error bars represent standard deviation). Bars with different letters indicate significant difference in multiple range comparison (one-way ANOVA followed by Tukey’s HSD test, p<0.05, n = 3).
DOI: http://dx.doi.org/10.7554/eLife.22502.005

Figure 2—figure supplement 2. The linkage disequilibrium pattern in the region surrounding DOG1.

Figure 2—figure supplement 2.

Manhattan plot of a local (same region as in Figure 2B–D) association scan for GR21 where the color scale indicate the extent of LD starting from the most strongly associated SNP (SNP1) at position 18,570,773. The r2 value between SNP1 and SNP2 is 0.61. The red box on the horizontal axis indicates the position of the DOG1 locus.
DOI: http://dx.doi.org/10.7554/eLife.22502.006

Figure 2—figure supplement 3. Details of local MLMM association scans for GR21.

Figure 2—figure supplement 3.

We used three different SNPs as starting point: SNP1 (A and D), SNP2 (B and E) or SNP3 (C and F). (A and C) Manhattan plots of local MLMM association scans. The previous and the next SNPs added as cofactors to the model are highlighted in light and dark orange, respectively, for each of the four forward regression steps. The DOG1 locus is marked by the red box on the horizontal axis. (D and F) Partition of variance at each step of MLMM (four forward and four backward) into variance explained by the SNPs included in the model (Explained), kinship (Genetic) and noise (Error). The vertical red bar shows the optimal number of steps according to the extended Bayesian information criteria (extBIC).
DOI: http://dx.doi.org/10.7554/eLife.22502.007