Table 1.

Overview of simulation settings to create synthetic phenotypes for real Arabidopsis thaliana genotypes.

Sim	n	k	y	c	ϵ
A	100	1	β 1x1 + u + ϵ	c 1 = 0.3	𝒩 (0, σ2)
B	500	1	β 1x1 + u + ϵ	c 1 = 0.3	𝒩 (0, σ2)
C	1,000	1	β 1x1 + u + ϵ	c 1 = 0.3	𝒩 (0, σ2)
D	2,000	1	β 1x1 + u + ϵ	c 1 = 0.3	𝒩 (0, σ2)
E	1,000	2	β 1x1 + β 2x2 + β 3x1 ◦ x2 + u + ϵ	c1 = c2 = 0.05, c 3 = 0.2	𝒩 (0, σ2)
F	1,000	2	β 1x1 + β 2x2 + β 3x1 ◦ x2 + u + ϵ	c 1 = c2 = 0.01, c3 = 0.28	𝒩 (0, σ2)
G	1,000	1	β 1x1 + u + ϵ	c 1 = 0.3	Γ(0, σ2/4)
H	1,000	1	β 1x1 + u + ϵ	c 1 = 0.3	Γ(0, σ2)
I	1,000	5	${\Sigma }_{i=1}^5$ β ixi + u + ϵ	ci ∼ 𝒩 (6, 22)/100, i = 1, . . . , 5	𝒩 (0, σ2)
J	1,000	20	${\Sigma }_{i=1}^{20}$ β ixi + u + ϵ	ci ∼ 𝒩 (1.5, 0.52)/100, i = 1, . . . , 20	𝒩 (0, σ2)
K	1,000	50	${\Sigma }_{i=1}^{50}$ β ixi + u + ϵ	ci ∼ 𝒩 (0.6, 0.22)/100, i = 1, . . . , 50	𝒩 (0, σ2)
L	1,000	100	${\Sigma }_{i=1}^{100}$ β ixi + u + ϵ	ci ∼ 𝒩 (0.3, 0.12)/100, i = 1, . . . , 100	𝒩 (0, σ2)

The first column Sim indicates the simulation setting, n is the number of samples, k the number of causal markers, y the formula of the phenotype, c the effect size of the causal markers, and ϵ the added noise. Here, x _i denotes a single marker and x _i ◦ x _j is the Hadamard product of two markers.