Table 3.
The Number of Interactions Identified from the WTCCC Data Sets of Seven Diseases under Different Constraints
| BD | CAD | CD | HT | RA | T1D | T2D | |
|---|---|---|---|---|---|---|---|
| C1 | 10 | 16 | 8 | 7 | 350 | 4499 | 18 |
| C1 & C2 | 0 | 0 | 1 | 0 | 0 | 789 | 0 |
| C1 & C2 & C3 | 0 | 0 | 1 | 0 | 0 | 91 | 0 |
Abbreviations are as follows: BD, bipolar disorder; CAD, coronary artery disease; CD, Crohn disease; HT, hypertension; RA, rheumatoid arthritis; T1D, type 1 diabetes; T2D, type 2 diabetes. C1 is the significance threshold constraint: the significance threshold is 0.05 for the Bonferroni-corrected interaction p value. C2 is the distance constraint: the physical distance between two interacting SNPs is at least 1Mb. This constraint is used to avoid interactions that might be attributed to the LD effects.4C3 is the main effect constraint: The single-locus p value should not be less than 10−6. This constraint is used to see whether there exist strong interactions without significant main effects, because those SNPs with p ≥ 10−6 are usually filtered out in the typical single-locus scan.