Table 1.
Characteristics of simulated covariate models
| Model | α | Inheritance model | Allele frequency | OR(G) | OR(E) | OR(GxE) | QTL | SD (QTL) | h2 (linked families) | h2 (dataset) | μL (SD) | μU (SD) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 0.2 | dominant | 0.10 | 1.0 | 33.1 | 1.0 | yes | 4 | 0.49 | 0.10 | – | – |
| 0.2 | recessive | 0.35 | 1.0 | 38.5 | 1.0 | yes | 4 | 0.40 | 0.08 | – | – | |
| 0.5 | dominant | 0.05 | 1.0 | 8.6 | 1.0 | yes | 4 | 0.35 | 0.07 | – | – | |
| 0.5 | recessive | 0.35 | 1.0 | 10.5 | 1.0 | yes | 4 | 0.40 | 0.08 | – | – | |
| 2 | 0.2 | dominant | 0.10 | 1.0 | 1.0 | 18.2 | no | – | – | – | – | – |
| 0.2 | recessive | 0.35 | 1.0 | 1.0 | 25.8 | no | – | – | – | – | – | |
| 0.5 | dominant | 0.05 | 1.0 | 1.0 | 8.0 | no | – | – | – | – | – | |
| 0.5 | recessive | 0.35 | 1.0 | 1.0 | 10.0 | no | – | – | – | – | – | |
| 3 | 0.2 | dominant | 0.10 | 10.5 | 1.0 | 1.0 | no | – | – | – | 40 (5) | 20 (5) |
| 0.2 | recessive | 0.35 | 12.8 | 1.0 | 1.0 | no | – | – | – | 40 (5) | 20 (5) | |
| 0.5 | dominant | 0.05 | 5.2 | 1.0 | 1.0 | no | – | – | – | 40 (5) | 20 (5) | |
| 0.5 | recessive | 0.35 | 6.0 | 1.0 | 1.0 | no | – | – | – | 40 (5) | 20 (5) | |
Constants: disease prevalence 5%, locus-specific λs = 1.15. α: proportion of families linked to analyzed marker map. λs for the entire dataset was calculated as the weighted average of λs in the linked subset of families (1.75 for α = 0.2, 1.30 for α = 0.5) and λs = 1 in the unlinked subset. For Model 1, genotype-specific co-variate means of 20, 30, and 30 were assumed for QTL genotypes aa, Aa, and AA in the dominant model, and means of 20, 20, and 30 were assumed in the recessive model, with the common standard deviation given in the SD(QTL) column. h2 is the theoretical heritability (additive and dominance variance component) that applies to the linked subset of families. h2 for the entire dataset was calculated as the weighted average of h2 in the linked subset of families and h2 = 0 in the unlinked subset. μL: covariate mean in linked families; μU: covariate mean in unlinked families. The ‘unlinked’ families (proportion 1-α), in which a second unlinked disease gene G2 was segregating, were generated with OR(G2) = 10, OR(E) = 1, RR(G2xE) = 1.