Limits...
Gene-based testing of interactions in association studies of quantitative traits.

Ma L, Clark AG, Keinan A - PLoS Genet. (2013)

Bottom Line: The framework of gene-based gene-gene interaction (GGG) tests combine marker-based interaction tests between all pairs of markers in two genes to produce a gene-level test for interaction between the two.Extensive simulations point to correct type I error rates of all tests and show that the two truncated tests are more powerful than the other tests in cases of markers involved in the underlying interaction not being directly genotyped and in cases of multiple underlying interactions.We conclude that our GGG tests show improved power to identify gene-level interactions in existing, as well as emerging, association studies.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, New York, USA. lm529@cornell.edu

ABSTRACT
Various methods have been developed for identifying gene-gene interactions in genome-wide association studies (GWAS). However, most methods focus on individual markers as the testing unit, and the large number of such tests drastically erodes statistical power. In this study, we propose novel interaction tests of quantitative traits that are gene-based and that confer advantage in both statistical power and biological interpretation. The framework of gene-based gene-gene interaction (GGG) tests combine marker-based interaction tests between all pairs of markers in two genes to produce a gene-level test for interaction between the two. The tests are based on an analytical formula we derive for the correlation between marker-based interaction tests due to linkage disequilibrium. We propose four GGG tests that extend the following P value combining methods: minimum P value, extended Simes procedure, truncated tail strength, and truncated P value product. Extensive simulations point to correct type I error rates of all tests and show that the two truncated tests are more powerful than the other tests in cases of markers involved in the underlying interaction not being directly genotyped and in cases of multiple underlying interactions. We applied our tests to pairs of genes that exhibit a protein-protein interaction to test for gene-level interactions underlying lipid levels using genotype data from the Atherosclerosis Risk in Communities study. We identified five novel interactions that are not evident from marker-based interaction testing and successfully replicated one of these interactions, between SMAD3 and NEDD9, in an independent sample from the Multi-Ethnic Study of Atherosclerosis. We conclude that our GGG tests show improved power to identify gene-level interactions in existing, as well as emerging, association studies.

Show MeSH

Related in: MedlinePlus

Average power of GGG tests summarized from Table 2.For each simulation scenario from Table 2, average power for each type of test is presented as an average across the different sample sizes (n) reported in Table 2. The method that collapses markers in each of the two genes, GG_PC, is least powerful in all simulation scenarios. Among the four GGG tests that combine P values, GG_minP and GG_GATES are more powerful only in simulation scenarios 3 and 4, which are the only cases that we simulated a single marker-by-marker interaction with both markers available for analysis (denoted by O-O in Table 2). GG_tTS and GG_tProd are most powerful in all other simulation scenarios.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3585009&req=5

pgen-1003321-g002: Average power of GGG tests summarized from Table 2.For each simulation scenario from Table 2, average power for each type of test is presented as an average across the different sample sizes (n) reported in Table 2. The method that collapses markers in each of the two genes, GG_PC, is least powerful in all simulation scenarios. Among the four GGG tests that combine P values, GG_minP and GG_GATES are more powerful only in simulation scenarios 3 and 4, which are the only cases that we simulated a single marker-by-marker interaction with both markers available for analysis (denoted by O-O in Table 2). GG_tTS and GG_tProd are most powerful in all other simulation scenarios.

Mentions: In all simulated scenarios, GG_PC, which takes the approach of first collapsing markers in each of the two genes, is less powerful than the four P value combining GGG tests (Table 2; Figure 2), which may be due to a combination of the principal components not fully capturing the underlying interaction signals and the multiple degrees of freedom associated with that test statistic. As both GG_minP and GG_GATES consider the best signal to represent a gene level interaction, they exhibit very similar levels of power, although GG_GATES is slightly more powerful in all simulated scenarios (Table 2; Figure 2). While GG_minP picks the smallest P value to represent a gene-level interaction, GG_GATES picks the strongest signal while accounting for the effective number of tests, which may not necessarily be the smallest P value, which explains the gain in power.


Gene-based testing of interactions in association studies of quantitative traits.

Ma L, Clark AG, Keinan A - PLoS Genet. (2013)

Average power of GGG tests summarized from Table 2.For each simulation scenario from Table 2, average power for each type of test is presented as an average across the different sample sizes (n) reported in Table 2. The method that collapses markers in each of the two genes, GG_PC, is least powerful in all simulation scenarios. Among the four GGG tests that combine P values, GG_minP and GG_GATES are more powerful only in simulation scenarios 3 and 4, which are the only cases that we simulated a single marker-by-marker interaction with both markers available for analysis (denoted by O-O in Table 2). GG_tTS and GG_tProd are most powerful in all other simulation scenarios.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3585009&req=5

pgen-1003321-g002: Average power of GGG tests summarized from Table 2.For each simulation scenario from Table 2, average power for each type of test is presented as an average across the different sample sizes (n) reported in Table 2. The method that collapses markers in each of the two genes, GG_PC, is least powerful in all simulation scenarios. Among the four GGG tests that combine P values, GG_minP and GG_GATES are more powerful only in simulation scenarios 3 and 4, which are the only cases that we simulated a single marker-by-marker interaction with both markers available for analysis (denoted by O-O in Table 2). GG_tTS and GG_tProd are most powerful in all other simulation scenarios.
Mentions: In all simulated scenarios, GG_PC, which takes the approach of first collapsing markers in each of the two genes, is less powerful than the four P value combining GGG tests (Table 2; Figure 2), which may be due to a combination of the principal components not fully capturing the underlying interaction signals and the multiple degrees of freedom associated with that test statistic. As both GG_minP and GG_GATES consider the best signal to represent a gene level interaction, they exhibit very similar levels of power, although GG_GATES is slightly more powerful in all simulated scenarios (Table 2; Figure 2). While GG_minP picks the smallest P value to represent a gene-level interaction, GG_GATES picks the strongest signal while accounting for the effective number of tests, which may not necessarily be the smallest P value, which explains the gain in power.

Bottom Line: The framework of gene-based gene-gene interaction (GGG) tests combine marker-based interaction tests between all pairs of markers in two genes to produce a gene-level test for interaction between the two.Extensive simulations point to correct type I error rates of all tests and show that the two truncated tests are more powerful than the other tests in cases of markers involved in the underlying interaction not being directly genotyped and in cases of multiple underlying interactions.We conclude that our GGG tests show improved power to identify gene-level interactions in existing, as well as emerging, association studies.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, New York, USA. lm529@cornell.edu

ABSTRACT
Various methods have been developed for identifying gene-gene interactions in genome-wide association studies (GWAS). However, most methods focus on individual markers as the testing unit, and the large number of such tests drastically erodes statistical power. In this study, we propose novel interaction tests of quantitative traits that are gene-based and that confer advantage in both statistical power and biological interpretation. The framework of gene-based gene-gene interaction (GGG) tests combine marker-based interaction tests between all pairs of markers in two genes to produce a gene-level test for interaction between the two. The tests are based on an analytical formula we derive for the correlation between marker-based interaction tests due to linkage disequilibrium. We propose four GGG tests that extend the following P value combining methods: minimum P value, extended Simes procedure, truncated tail strength, and truncated P value product. Extensive simulations point to correct type I error rates of all tests and show that the two truncated tests are more powerful than the other tests in cases of markers involved in the underlying interaction not being directly genotyped and in cases of multiple underlying interactions. We applied our tests to pairs of genes that exhibit a protein-protein interaction to test for gene-level interactions underlying lipid levels using genotype data from the Atherosclerosis Risk in Communities study. We identified five novel interactions that are not evident from marker-based interaction testing and successfully replicated one of these interactions, between SMAD3 and NEDD9, in an independent sample from the Multi-Ethnic Study of Atherosclerosis. We conclude that our GGG tests show improved power to identify gene-level interactions in existing, as well as emerging, association studies.

Show MeSH
Related in: MedlinePlus