tProd.test: Gene-based Gene-Gene Interaction analysis by combining...

In GeneGeneInteR: Tools for Testing Gene-Gene Interaction at the Gene Level

Description

gates.test, minP.test, tTS.test and tProd.test aim at performing gene-gene interaction analysis based on SNP-SNP interaction tests. The following methods are used to combine SNP-SNP interaction tests into a single Gene-Gene Interaction p-value:

• Minimum p-value in minP.test function

• Gene Association Test with Extended Simes procedure in gates.test

• Truncated Tail Strength in tTS.test function

• Truncated p-value Product in tProd.test function

Usage

tProd.test(Y, G1, G2, tau = 0.05, n.sim = 1000)

Arguments

Y	numeric or factor vector with exactly two different values. Y is the response variable and should be of length equal to the number of rows of G1 and G2 arguments (number of individuals).
G1	SnpMatrix object. Must have a number of rows equal to the length of Y.
G2	SnpMatrix object. Must have a number of rows equal to the length of Y.
tau	numeric in [0, 1]. See details section for its use.
n.sim	positive integer. n.sim is the number of multivariate normal distribution simulations used to compute the significant level for tTS and tProd methods.

Details

In a first step, all methods start by applying a logistic regression model to test all pairs of SNPs between the two genes G1 and G2. If G1 has m1 SNPs and G2 m2 SNPs, a total of m1 * m2 SNP-SNP tests are performed. In a second step, the m1 * m2 SNP-SNP tests are combined according to their covariance matrix Σ. Σ is computed as described in the method developped in Emily (2016). The covariance Σ is used in each method as follows:

• minP test - minP test considered the significant of the observed minimum p-value. Significance is computed by integrating the multivariate normal distribution with covariance Σ as proposed in Conneelly and Boehnke (2008).

• GATES test - The p-value for GATES is the minimum p-value obtained after a multiple testing correction of the SNP-SNP interaction p-values. Correction for multiple testing is defined as

  me * p[i]/me[i]

  where me is the effective number of independant tests, p[i] is the i-th top p-values and me[i] is the effective number of independant test among the top i p-values. Many methods exist to estimate me and me[i] terms:

  • Cheverud-Nyholt method (Cheverud, 2001 and Nyholt, 2004)

  • Keff method (Moskovina and Schmidt, 2008)

  • Li & Ji method (Li and Ji, 2005)

  • Galwey method (Galwey, 2009)

  Details of each method can be found in the references.

• tTS test - tTS test does not consider only the strongest signal but all signals that are inferior to a given threshold τ. For these p-values, the weighted sum of

  tTS=∑ (1-p[i]*(m1*m2+1)/i)

  is computed and represents the test statistic. The p-value is calculated using an empirical distribution of tTS obtained by simulating multivariate normal statistics with a covariance Σ as proposed by Jiang et al. (2011).

• TProd test - The procedure is similar to tTS with tProd defined as

  tProd=∏ p[i].

  See Zaykin et al. (2002) for details.

Value

A list with class "GGItest" containing the following components:

statistic	The value of the statistic tProd.
p.value	The p-value for the test
estimate	Estimation of tProd.
parameter	The threshold value tau.
alternative	a character string describing the alternative.
method	a character string indicating the method used.
data.name	a character string giving the names of the data.

References

M. Emily AGGrEGATOr: A Gene-based GEne-Gene interActTiOn test for case-control association studies, Statistical Application in Genetics and Molecular Biology, 15(2): 151-171, 2016.
L. Ma, A.G. Clark and A. Keinan Gene-Based Testing Of Interactions in Association Studies of Quantitative Traits. PLoS Genetics 9(2):e1003321, 2013.
V. Moskvina and K.M. Schmidt On multiple-testing correction in genome-wide association studies. Genetic Epidemiology, 32(6): 567-573, 2008.
J. Li and L. Ji. Adjusting multiple testing in multilocus analyses using the eigenvalues of a correlation matrix. Heredity 95: 221-227, 2005.
N.W. Galwey. A new measure of the effective number of tests, a practical tool for comparing families of non-independent significance tests. Genetic Epidemiology 33(7): 559-568, 2009.
J.M. Cheverud. A simple correction for multiple comparisons in interval mapping genome scans. Heredity. 87(1):52-58, 2001.
D.R. Nyholt. A Simple Correction for Multiple Testing for Single-Nucleotide Polymorphisms in Linkage Disequilibrium with Each Other. American journal of human genetics. 74(4): 765-769, 2004.
K.N. Conneely and M. Boehnke. So many correlated tests, so little time! rapid adjustment of p values for multiple correlated tests. The American Journal of Human Genetics, 81: 1158-1168, 2008.
B. Jiang, X. Zhang, Y. Zuo and G. Kang. A powerful truncated tail strength method for testing multiple null hypotheses in one dataset. Journal of Theoretical Biology 277: 67-73, 2011.
D.V. Zaykin, L.A. Zhivotovsky, P.H. Westfall and B.S. Weir. Truncated product method for combining P-values. Genetic epidemiology 22: 170-185, 2002.

See Also

GGI

Examples

data(gene.pair)
tProd.test(gene.pair$Y, gene.pair$G1, gene.pair$G2, tau = 0.5, n.sim = 500)

GeneGeneInteR documentation built on Nov. 8, 2020, 6:28 p.m.