Table 4.
Analysis of two-way SNP interaction effects on breast cancer.
| Two-way Interactions between polymorphisms | Crude P-values‡ | Bootstrap frequency of stepwise variable selection* | Bootstrap P-values† | FDR adjusted P-values§ | FPRP¶ |
| XPD-[Lys751Gln] and IL10-[G(-1082)A] | 0.035 | 68% | 0.001 | 0.007 | 0.092 |
| COMT-[Met108/158Val] and CCND1-[Pro241Pro] | 0.010 | 61% | 0.002 | 0.014 | 0.169 |
| GSTP1-[Ile105Val] and COMT-[Met108/158Val] | 0.036 | 54% | 0.001 | 0.007 | 0.093 |
| CYP17-[C(518)T] and GADD45-[C(IVS3+168)T] | 0.024 | 53% | 0.018 | 0.062 | 0.999 |
| BARD1-[Pro24Ser] and ESR1-[Pro325Pro] | 0.039 | 51% | ns|| | ns | - |
| TNFA-[G(-308)A] and p27-[Val109Gly] | 0.016 | 49% | 0.025 | 0.079 | 0.996 |
| BARD1-[Pro24Ser] and p27-[Val109Gly] | 0.021 | 44% | ns | ns | - |
| BARD1-[Pro24Ser] and XPD-[Lys751Gln] | 0.024 | 36% | 0.002 | 0.014 | 0.671 |
| ESR1-[Ser10Ser] and ESR1-[Pro325Pro] | 0.028 | 30% | 0.097 | ns | 0.999 |
‡ Probability to enter the model in the stepwise multivariable model based on the score statistic in the original data set.
*Proportion of times the interaction is selected by the stepwise multivariable model in the 1,000 bootstrap samples.
†P-values estimated by bootstrap analysis in the 1,000 random samples.
§False Discovery Rate (FDR) adjusted P-values.
||ns: p > 0.10.
¶False Positive Report Probability