epiGWAS (version 1.0.2)

sample_SNP: Samples causal SNPs with different effect types

Description

The sampled SNPs are combined in a list of character vectors with the following fields: target, marginal, inter1 and inter2. Through the parameters overlap_marg and overlap_inter, the synergistic SNPs with the target can have additional marginal and epistatic effects. The SNPs are consecutively sampled in the following order: target, marginal, inter1 and inter2. For each SNP, we iteratively sample until the picked SNP candidate meets the constraints defined by thresh_MAF and window_size (see arguments for more details) or until the maximum number of resamplings is reached. To avoid duplication of effects, we sample at most one SNP per cluster.

Usage

sample_SNP(nX, nY, nZ12, clusters, MAF, thresh_MAF = 0.2,
  window_size = 3, overlap_marg = 0, overlap_inter = 0,
  max_iter = 10000)

Arguments

nX

number of SNPs interacting with the target variant

nY

number of SNPs with marginal effects

nZ12

number of SNP pairs with epistatic effects

clusters

vector of cluster memberships. Typically, the output of cutree. For ease of identification, the SNP IDs in names(clusters) are mandatory.

MAF

vector of minor allele frequencies. The order of the SNPs in MAF is identical to that in clusters.

thresh_MAF

lower-bound on the minor allele frequencies of causal SNPs. Rare variants are inherently difficult to recover. Assessing the retrieval performance on common variants better reflects the true performance of the epistasis detection algorithm.

window_size

in number of clusters. Beside the target variant, the causal SNPs are sampled outside of a window centered around the target. On each side of the target variant, the width of the window is window_size.

overlap_marg

number of SNPs with both synergistic effects with the target and marginal effects

overlap_inter

number of SNPs with both synergistic effects with the target and additional epistatic effects

max_iter

maximum number of sampling rejections for each SNP. If exceeded, the function generates an error

Value

list of character vectors containing to the causal SNP IDs. The output list entries are: target, marginal, inter1 and inter2. An epistatic pair is obtained from the combination of two SNPs with identical positions in inter1 and inter2.

Warning

Make sure to supply the SNP IDs in names(clusters). The SNPs in the output list are referenced by their names.

Examples

# NOT RUN {
clusters <- rep(seq_len(10), each = 3)
names(clusters) <- paste0("SNP_", seq_along(clusters))
MAF <- runif(length(clusters), min = 0.1, max = 0.5)

sample_SNP(nX = 2, nY = 2, nZ12 = 1, clusters, MAF, thresh_MAF = 0.2,
           window_size = 2, overlap_marg = 1, overlap_inter = 0)

# }