Limits...
Power comparison of different methods to detect genetic effects and gene-environment interactions.

Kazma R, Dizier MH, Guilloud-Bataille M, Bonaïti-Pellié C, Génin E - BMC Proc (2007)

Bottom Line: However, their respective performances have rarely been compared.Results show that for the particular model of interaction between tobacco use and Locus B simulated here, the mean interaction test has poor power to detect either the genetic effect or the interaction.The association studies, i.e., the log-linear-modeling approach and the case-control method, are more powerful to detect the genetic effect (power of 78% and 95%, respectively) and taking into account interaction moderately increases the power (increase of 9% and 3%, respectively).

View Article: PubMed Central - HTML - PubMed

Affiliation: Université Paris-Sud, UMR-S 535, Villejuif, 94817, France. kazma@vjf.inserm.fr

ABSTRACT
Identifying gene-environment (G x E) interactions has become a crucial issue in the past decades. Different methods have been proposed to test for G x E interactions in the framework of linkage or association testing. However, their respective performances have rarely been compared. Using Genetic Analysis Workshop 15 simulated data, we compared the power of four methods: one based on affected sib pairs that tests for linkage and interaction (the mean interaction test) and three methods that test for association and/or interaction: a case-control test, a case-only test, and a log-linear approach based on case-parent trios. Results show that for the particular model of interaction between tobacco use and Locus B simulated here, the mean interaction test has poor power to detect either the genetic effect or the interaction. The association studies, i.e., the log-linear-modeling approach and the case-control method, are more powerful to detect the genetic effect (power of 78% and 95%, respectively) and taking into account interaction moderately increases the power (increase of 9% and 3%, respectively). The case-only design exhibits a 95% power to detect G x E interaction but the type I error rate is increased.

No MeSH data available.


Difference in p-values of G+I and G tests. Difference is represented for the case-control (red plot) and the log-linear-modeling (blue plot) by ln(pG)-ln(pG+I) reported over the first 25 replicates.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2367459&req=5

Figure 1: Difference in p-values of G+I and G tests. Difference is represented for the case-control (red plot) and the log-linear-modeling (blue plot) by ln(pG)-ln(pG+I) reported over the first 25 replicates.

Mentions: With the log-linear model, the power to detect the gene effect is 78% and is increased to 87% when accounting for G × E interaction. Thus, there is a gain in power to detect the gene effect when accounting for G × E interaction under the simulated model. For the case-control design, the power to detect the gene effect is 95% and improves to 98% when accounting for interaction. As shown in Figure 1, the p-values of test accounting for G × E are smaller than those of the test not accounting for G × E for most of the replicates and similar trends (gain or loss of power) are observed between the two methods in 74% of the replicates.


Power comparison of different methods to detect genetic effects and gene-environment interactions.

Kazma R, Dizier MH, Guilloud-Bataille M, Bonaïti-Pellié C, Génin E - BMC Proc (2007)

Difference in p-values of G+I and G tests. Difference is represented for the case-control (red plot) and the log-linear-modeling (blue plot) by ln(pG)-ln(pG+I) reported over the first 25 replicates.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2367459&req=5

Figure 1: Difference in p-values of G+I and G tests. Difference is represented for the case-control (red plot) and the log-linear-modeling (blue plot) by ln(pG)-ln(pG+I) reported over the first 25 replicates.
Mentions: With the log-linear model, the power to detect the gene effect is 78% and is increased to 87% when accounting for G × E interaction. Thus, there is a gain in power to detect the gene effect when accounting for G × E interaction under the simulated model. For the case-control design, the power to detect the gene effect is 95% and improves to 98% when accounting for interaction. As shown in Figure 1, the p-values of test accounting for G × E are smaller than those of the test not accounting for G × E for most of the replicates and similar trends (gain or loss of power) are observed between the two methods in 74% of the replicates.

Bottom Line: However, their respective performances have rarely been compared.Results show that for the particular model of interaction between tobacco use and Locus B simulated here, the mean interaction test has poor power to detect either the genetic effect or the interaction.The association studies, i.e., the log-linear-modeling approach and the case-control method, are more powerful to detect the genetic effect (power of 78% and 95%, respectively) and taking into account interaction moderately increases the power (increase of 9% and 3%, respectively).

View Article: PubMed Central - HTML - PubMed

Affiliation: Université Paris-Sud, UMR-S 535, Villejuif, 94817, France. kazma@vjf.inserm.fr

ABSTRACT
Identifying gene-environment (G x E) interactions has become a crucial issue in the past decades. Different methods have been proposed to test for G x E interactions in the framework of linkage or association testing. However, their respective performances have rarely been compared. Using Genetic Analysis Workshop 15 simulated data, we compared the power of four methods: one based on affected sib pairs that tests for linkage and interaction (the mean interaction test) and three methods that test for association and/or interaction: a case-control test, a case-only test, and a log-linear approach based on case-parent trios. Results show that for the particular model of interaction between tobacco use and Locus B simulated here, the mean interaction test has poor power to detect either the genetic effect or the interaction. The association studies, i.e., the log-linear-modeling approach and the case-control method, are more powerful to detect the genetic effect (power of 78% and 95%, respectively) and taking into account interaction moderately increases the power (increase of 9% and 3%, respectively). The case-only design exhibits a 95% power to detect G x E interaction but the type I error rate is increased.

No MeSH data available.