AFpca: Adaptive Fisher Method with Principal Components (AFpca) for...
In cxystat/AFb: B-spline Based Adaptive Fisher Method

Description Usage Arguments Value Examples

Adaptive Fisher Method with Principal Components (AFpca) for Trait-Methylation Set Association

AFpca(
  Y,
  M,
  binary = FALSE,
  cov = NULL,
  varprop = 0.95,
  nperm = 1000,
  n0 = 1,
  adapt_perm = FALSE,
  cutoff = 2.5e-06,
  seed = NULL
)

`Y`	Y Phenotype data. It can be a continuous trait or a binary trait. A vector of length `n` (number of subjects).
`M`	A matrix of methylation levels with dimensions `n` by `K` (`n` subjects, `K` CpG sites).
`binary`	Indicator of whether `Y` is binary.
`cov`	Covariates. A matrix with dimensions `n` by `J` (number of covariates).
`varprop`	Cutoff for proportion of variance to be expalined. The first Kp principal components (PCs) are chosen such that `varprop` of the total variance is explained.
`nperm`	Number of permutations. Also the starting number of permutations for "step-up" algorithm. Default is `1,000`.
`n0`	Tuning parameter. Discard the first `n0-1` P-values of each sample.
`adapt_perm`	Whether "step-up" algorithm is used for P-value calculation. If FALSE, function permutes `nperm` times and stops. If TRUE, `nperm` will be increased `10` times each round if P-value <= `5/nperm`. Algorithm stops if P-value > `5/nperm` or <= `cutoff`.
`cutoff`	Cutoff for "step-up" algorithm.
`seed`	Specify the seed for permutations.

An object of "AFpca" class.

pv: P-value of AFb test.
stat: Test statistic of AFb test.
indexes: Indexes of PCs combined into the test statistic. Indexes are sorted so that P-values are in ascending order.
stat_all: AFb statistics for all permuted samples.
pv_all: P-values of AFb statistics for all permuted samples.
method: Method used.

Y <- bs_dense$trait
methyl <- bs_dense$methyl
pos <- bs_dense$pos
test <- AFpca(Y, methyl, varprop = 0.9,
            binary = TRUE, adapt_perm = TRUE)
summary(test)