TABLE 1.
Posterior probability |
||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Risk set | SNP8 | SNP16 | SNP27 | SNP34 | SNP46 | SNP53 | SNP77 | SNP82 | N | Mean | General | Specific |
1 | HE/HO | HE/HO | 447–587 | 103.1 (102.3–103.9) | 0.97 | 0.72 | ||||||
2 | HE/HO | HE/HO | HE/HO | 317–380 | 103.7 (102.8–104.6) | 0.69 | 0.47 | |||||
3 | HE/HO | HE/HO | HE/HO | 196–267 | 104.2 (103.0–105.4) | 0.09 | 0.05 | |||||
4 | HE | 376–2053 | 100.6 (100.1–101.3) | 0.14 | 0.06 | |||||||
5 | HE/HO | 166–872 | 101.1 (100.2–102.1) | 0.08 | 0.04 |
N, the number of individuals in the risk group; HE, heterozygote; HO, homozygote; mean, the posterior estimate of the mean phenotype of the risk group. Results are shown for the simulated scenario described in Figure 3. Posterior probabilities were calculated as the probability of a parameter combination being part of a risk set (general) and the probability of a combination being part of a risk set excluding combinations where other parameters are also included in the risk set (specific). Large differences between the general and specific posterior probability mean that other parameters are important for this group of risk sets. For illustration of risk set grouping all combinations with a general posterior probability of >0.05 are included. Because of a low number of homozygotes some of the risk sets are very similar and are manually grouped together so that the SNP parameters (HE or HO) and (HE) were grouped together. Although this method performs no formal test for interaction, interactions are suggested when some parameters are sampled only in combination and rarely alone. The number of individuals in a risk set and the mean for each risk set are given as 90% credibility intervals. The nonsusceptibility SNPs SNP16, SNP27, and SNP53 are in LD with the true susceptibility SNPs SNP8, SNP34, and SNP46 respectably (D′ > 0.98, r2 < 0.7).