effectPDF: Genetic interaction testing given effect size distribution
In WeiAkaneDeng/GEWIST: Gene Environment Wide Interaction Search Threshold

Description Usage Arguments Value Warning Author(s) References Examples

Compute the optimal Variance Prioritization power and corresponding Levene's test p-value threshold for prioritization given the interaction effect size distribution using GEWIST.

1
2
3

effectPDF(distribution = c("beta", "normal", "uniform", "weibull"),
parameter1, parameter2 = NULL, parameter3 = NULL, p, N, theta_c, M, 
K = 20000, nb_incr = 50,  range = NULL, verbose = FALSE)

`distribution`	distribution of interaction effect size. Possible distributions are: "beta" for beta distribution "normal" for normal distribution "uniform" for uniform distribution "weibull" for weibull distribution
`parameter1`	the first parameter used in the corresponding distribution
`parameter2`	the second parameter used in the corresponding distribution, could set to be null
`parameter3`	the third parameter used in the corresponding distribution, could set to be null
`p`	minor allele frequency of the SNP, a number between 0 and 0.5
`N`	sample size
`theta_c`	proportion of quantitative trait variance explained by the covariate, should be a number between 0 and 1
`M`	total number of SNPs to be tested
`K`	number of GEWIST procedures, by default, set to be 20,000
`nb_incr`	number of effect size points in the range to be prioritized using GEWIST; by default set to be 50.
`range`	range of variance explained by interaction effect sizes, a vector of length 2
`verbose`	logical; if TRUE, for each interaction effect size, function returns a data.frame class object listing the VP power at each p-value, from 0.001 to 1 with 0.001 incremental increase.

A list with three components:

`Optimal_VP_power`	VP power to detect interactions at the optimal VP p-value threshold
`Conventional_power`	power to detect interactions without prioritization, i.e, VP power at Levene' test p-value of 1
`Optimal_pval_threshold`	levene'e test p-value at which optimal VP power is achieved

Computational time is directly proportional to nb_incr.

Wei Q. Deng <dengwq@mcmaster.ca> Guillaume Pare <pareg@mcmaster.ca>

Deng W.Q, Pare G. (2011) A fast algorithm to optimize SNP prioritization for gene-gene and gene-environment interactions. Genetic Epidemiology. 35: 729-738. doi: 10.1002/gepi.20624

Pare G, Cook NR, Ridker PM, Chasman DI (2010) On the Use of Variance per Genotype as a Tool to Identify Quantitative Trait Interaction Effects: A Report from the Women's Genome Health Study. PLoS Genet 6(6): e1000981. doi:10.1371/journal.pgen.1000981

Levene H. (1960) Robust tests for equality of variances. In Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling eds:I. Olkin, S.G. Ghurye, W. Hoeffding, W.G. Madow & H.B.Mann, pp.278-292. Stanford: Stanford University Press.

# Given a SNP with minor allele frequency of 10% and a sample 
# of 10,000 individuals, we are interested in testing interactions 
# between this SNP and a covariate of effect size 10%. The 
# total number of SNP is 500,000. Assume the unknown interaction 
# effect size has a Weibull distribution in the range of 0.05% 
# and 0.3% variance explained with 50 increments.Repeat GEWIST 
# for each of the 50 interaction effect sizes.  


library(GEWIST)
effectPDF(distribution = "weibull", parameter1 = 0.8, parameter2 = 0.3,
parameter3 = NULL, p = 0.1 ,N = 10000, theta_c = 0.1, M = 350000,
K = 20000, nb_incr = 50, range = c(0.05/100,0.3/100), verbose = FALSE)

## End of script