Limits...
Practical and theoretical considerations in study design for detecting gene-gene interactions using MDR and GMDR approaches.

Chen GB, Xu Y, Xu HM, Li MD, Zhu J, Lou XY - PLoS ONE (2011)

Bottom Line: To provide empirical guidelines for planning such studies and data analyses, we investigated the performance of the multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) methods under various experimental scenarios.However, the two methods performed similarly when the accuracy was outside this range or the sample was significantly larger.We conclude that with adjustment of a covariate, GMDR performs better than MDR and a sample size of 1000∼2000 is reasonably large for detecting gene-gene interactions in the range of effect size reported by the current literature; whereas larger sample size is required for more subtle interactions with accuracy <0.56.

View Article: PubMed Central - PubMed

Affiliation: Institute of Bioinformatics, Zhejiang University, Hangzhou, Zhejiang, People's Republic of China.

ABSTRACT
Detection of interacting risk factors for complex traits is challenging. The choice of an appropriate method, sample size, and allocation of cases and controls are serious concerns. To provide empirical guidelines for planning such studies and data analyses, we investigated the performance of the multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) methods under various experimental scenarios. We developed the mathematical expectation of accuracy and used it as an indicator parameter to perform a gene-gene interaction study. We then examined the statistical power of GMDR and MDR within the plausible range of accuracy (0.50∼0.65) reported in the literature. The GMDR with covariate adjustment had a power of >80% in a case-control design with a sample size of ≥2000, with theoretical accuracy ranging from 0.56 to 0.62. However, when the accuracy was <0.56, a sample size of ≥4000 was required to have sufficient power. In our simulations, the GMDR outperformed the MDR under all models with accuracy ranging from 0.56∼0.62 for a sample size of 1000-2000. However, the two methods performed similarly when the accuracy was outside this range or the sample was significantly larger. We conclude that with adjustment of a covariate, GMDR performs better than MDR and a sample size of 1000∼2000 is reasonably large for detecting gene-gene interactions in the range of effect size reported by the current literature; whereas larger sample size is required for more subtle interactions with accuracy <0.56.

Show MeSH
Asymptotic trends of testing accuracy with different sample sizes.The result was based on a checkerboard model whose parameters were the same as shown in Figure 1. The solid lines are the analytical accuracy and represent the upper bound of the testing accuracy. The three lines downward are the means of the testing accuracies from 200 simulations with a sample size of 2000, 1000, and 500. Because the lines for a sample size of>2000 are coincident with the analytical accuracy lines, they are not shown.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3046176&req=5

pone-0016981-g002: Asymptotic trends of testing accuracy with different sample sizes.The result was based on a checkerboard model whose parameters were the same as shown in Figure 1. The solid lines are the analytical accuracy and represent the upper bound of the testing accuracy. The three lines downward are the means of the testing accuracies from 200 simulations with a sample size of 2000, 1000, and 500. Because the lines for a sample size of>2000 are coincident with the analytical accuracy lines, they are not shown.

Mentions: TA is commonly used in GMDR and MDR. Because it is context-dependent, its mathematical expectation is difficult to derive straightforwardly. Empirically, we show in Figure 2 that when the sample size increases to infinity under a checkerboard model, TA approaches accuracy, which is the theoretical upper bound of TA. For the cases illustrated, TA closely converges to accuracy with a sample size of 1000∼2000. The upper limit of TA can be attained when, in the testing set, each genotypic cell is recognized correctly as high or low risk after the cell has been classified correctly in the training set.


Practical and theoretical considerations in study design for detecting gene-gene interactions using MDR and GMDR approaches.

Chen GB, Xu Y, Xu HM, Li MD, Zhu J, Lou XY - PLoS ONE (2011)

Asymptotic trends of testing accuracy with different sample sizes.The result was based on a checkerboard model whose parameters were the same as shown in Figure 1. The solid lines are the analytical accuracy and represent the upper bound of the testing accuracy. The three lines downward are the means of the testing accuracies from 200 simulations with a sample size of 2000, 1000, and 500. Because the lines for a sample size of>2000 are coincident with the analytical accuracy lines, they are not shown.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3046176&req=5

pone-0016981-g002: Asymptotic trends of testing accuracy with different sample sizes.The result was based on a checkerboard model whose parameters were the same as shown in Figure 1. The solid lines are the analytical accuracy and represent the upper bound of the testing accuracy. The three lines downward are the means of the testing accuracies from 200 simulations with a sample size of 2000, 1000, and 500. Because the lines for a sample size of>2000 are coincident with the analytical accuracy lines, they are not shown.
Mentions: TA is commonly used in GMDR and MDR. Because it is context-dependent, its mathematical expectation is difficult to derive straightforwardly. Empirically, we show in Figure 2 that when the sample size increases to infinity under a checkerboard model, TA approaches accuracy, which is the theoretical upper bound of TA. For the cases illustrated, TA closely converges to accuracy with a sample size of 1000∼2000. The upper limit of TA can be attained when, in the testing set, each genotypic cell is recognized correctly as high or low risk after the cell has been classified correctly in the training set.

Bottom Line: To provide empirical guidelines for planning such studies and data analyses, we investigated the performance of the multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) methods under various experimental scenarios.However, the two methods performed similarly when the accuracy was outside this range or the sample was significantly larger.We conclude that with adjustment of a covariate, GMDR performs better than MDR and a sample size of 1000∼2000 is reasonably large for detecting gene-gene interactions in the range of effect size reported by the current literature; whereas larger sample size is required for more subtle interactions with accuracy <0.56.

View Article: PubMed Central - PubMed

Affiliation: Institute of Bioinformatics, Zhejiang University, Hangzhou, Zhejiang, People's Republic of China.

ABSTRACT
Detection of interacting risk factors for complex traits is challenging. The choice of an appropriate method, sample size, and allocation of cases and controls are serious concerns. To provide empirical guidelines for planning such studies and data analyses, we investigated the performance of the multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) methods under various experimental scenarios. We developed the mathematical expectation of accuracy and used it as an indicator parameter to perform a gene-gene interaction study. We then examined the statistical power of GMDR and MDR within the plausible range of accuracy (0.50∼0.65) reported in the literature. The GMDR with covariate adjustment had a power of >80% in a case-control design with a sample size of ≥2000, with theoretical accuracy ranging from 0.56 to 0.62. However, when the accuracy was <0.56, a sample size of ≥4000 was required to have sufficient power. In our simulations, the GMDR outperformed the MDR under all models with accuracy ranging from 0.56∼0.62 for a sample size of 1000-2000. However, the two methods performed similarly when the accuracy was outside this range or the sample was significantly larger. We conclude that with adjustment of a covariate, GMDR performs better than MDR and a sample size of 1000∼2000 is reasonably large for detecting gene-gene interactions in the range of effect size reported by the current literature; whereas larger sample size is required for more subtle interactions with accuracy <0.56.

Show MeSH