gt.object-class: Class "gt.object" for storing the result of the function gt

In globaltest: Testing Groups of Covariates/Features for Association with a Response Variable, with Applications to Gene Set Testing

Description

The class gt.object is the output of a call to gt. It stores the information needed for various diagnostic plots.

Slots

These slots are not meant to be directly accessed by the user.

result:

Object of class "matrix". The number of rows of this matrix is the number of tests performed. The matrix has at least the columns "p-value", "Statistic" "Expected", "Std.dev", and "#Cov".

extra:

Object of class "data.frame". Holds additional information that may be added later about the tests performed, such as multiplicity-adjusted p-values (see p.adjust), alias names for tests and comparative proportions (see comparative).

call:

The matched call to gt.

functions:

A "list" of various functions used by the covariates and subjects functions and the various methods.

subsets:

A "list" or "NULL". Stores the subsets tested, if more than one.

structure:

A "list" or "NULL". Stores subset and superset relationships between the sets in the "subsets" slot.

weights:

A "list" or "NULL". Stores the weight vectors used for testing, if more than one.

alternative:

If gt was called with x = TRUE, stores the design matrix of the alternative hypothesis; "NULL" otherwise.

null:

If gt was called with x = TRUE, stores the design matrix of the null hypothesis; "NULL" otherwise.

directional

Stores the directional argument of the call to gt.

legend

Object of class "list". Stores appropriate legends for the covariates and subjects plots.

model

Object of class "character". Stores the model.

Methods

show

(gt.object): Prints the test results: p-value, test statistic, expected value of the test statistic under the null hypothesis, standard deviation of the test statistic under the null hypothesis, and number of covariates tested.

summary

(gt.object): Prints the test results (as show) plus additional information on the model and the test.

p.value

(gt.object): Extracts the p-values.

z.score

(gt.object): Extracts z-score: (Test statistic - Expected value) / Standard deviation.

result

(gt.object): Extracts the results matrix together with the additional (e.g. multiple testing) information in the extra slot.

extract

(gt.object): Extracts the results matrix for the leaf nodes after a call to link{covariates}, with information on direction of association.

sort

(gt.object): Sorts the pathways to increasing p-values. Equal p-values are sorted on decreasing z-scores.

"["

(gt.object): Extracts results of one or more test results if multiple tests were performed. Identical to "[[".

"[["

(gt.object): Extracts results of one or more test results if multiple tests were performed. Identical to "[".

length

(gt.object): The number of tests performed.

size

(gt.object): Extracts a vector with the number of alternative covariates tested for each test.

names

(gt.object): Extracts the row names of the results matrix.

names<-

(gt.object): Changes the row names of the results matrix. Duplicate names are not allowed, but see alias.

alias

(gt.object): Extracts the "alias" column of the results matrix that can be used to add additional information on each test perfomed.

alias<-

(gt.object): Changes the "alias" column of the results matrix. Note that unlike for names, duplicate aliases are allowed.

weights

(gt.object): extracts the effective weights of the covariates as they are used internally by the test.

subsets

(gt.object): extracts the "subsets" slot.

hist

(gt.object): Produces a histogram to visualize the permutation test statistics. Only relevant after permutation testing.

covariates

(gt.object): Produces a plot to show the influence of individual covariates on the test result. See covariates for details.

subjects

(gt.object): Produces a plot to show the influence of individual subjects on the test result. See subjects for details.

p.adjust

(gt.object): Performs multiple testing correction and produces multiplicity-corrected p-values. See p.adjust for details.

comparative

(gt.object): Compares the p-values of tests performed on a subsets or weights with p-values of random subsets of covariates of same size or randomly distributed weights. See comparative for details.

sterms

(gt.object): Prints the smooth terms specified by gtPS, gtKS or gtLI.

Author(s)

Jelle Goeman: [e-mail]; Jan Oosting

See Also

gt, covariates, subjects.

Examples

    # Simple examples with random data here
    # Real data examples in the Vignette

    # Random data: covariates A,B,C are correlated with Y
    Y <- rnorm(20)
    X <- matrix(rnorm(200), 20, 10)
    X[,1:3] <- X[,1:3] + 0.5*Y
    colnames(X) <- LETTERS[1:10]

    # Make a gt.object
    sets <- list(odd = c(1,3,5,7,9), even = c(2,4,6,8,10))
    res <- gt(Y, X, subsets=sets)

    # Show the results
    res
    summary(res)
    sort(res)
    p.value(res)
    subsets(res)

    # Names
    names(res)
    names(res) <- c("ODD", "EVEN")
    alias(res) <- c("odd covariates", "even covariates")

    # Multiple testing
    p.adjust(res, method = "holm")
    p.adjust(res, method = "BH")

    # Diagnostics
    weights(res)
    covariates(res[1])
    extract(covariates(res[1]))
    subjects(res[1])

    # Permutation testing
    res <- gt(Y, X, perm = 1e4)
    hist(res)

globaltest documentation built on Nov. 8, 2020, 8:18 p.m.