ATC {cola}R Documentation

Ability to correlate other rows in the matrix

Description

Ability to correlate other rows in the matrix

Usage

ATC(mat, cor_fun = stat::cor, min_cor = 0, power = 1,
    mc.cores = 1, n_sampling = 1000, q_sd = 0, group = NULL, ...)

Arguments

mat

A numeric matrix. ATC score is calculated by rows.

cor_fun

A function which calculates correlations.

min_cor

Cutoff for the minimal absolute correlation.

power

Power on the correlation values.

mc.cores

Mumber of cores.

n_sampling

When there are too many rows in the matrix, to get the curmulative distribution of how one row correlates other rows, actually we don't need to use all the rows in the matrix, e.g. 1000 rows can already give a very nice estimation.

q_sd

Percentile of the standard deviation for the rows. Rows with values less than it are ignored.

group

A categorical variable. If it is specified, the correlation is only calcualted for the features in the same group.

...

Pass to cor_fun.

Details

For a given row in a matrix, the ATC score is the area above the curve of the curmulative density distribution of the absolute correlation to all other rows. Formally, if F_i(X) is the cumulative distribution function of X where X is the absolute correlation for row i with power power (i.e. x = cor^power), ATC_i = 1 - \int_{min_cor}^1 F_i(X).

By default the ATC scores are calculated by Pearson correlation, to use Spearman correlation, you can register the top-value method by:

    register_top_value_methods(
        "ATC_spearman" = function(m) ATC(m, method = "spearman")
    )  

Similarly, to use a robust correlation method, e.g. bicor function, you can do like:

    register_top_value_methods(
        "ATC_bicor" = function(m) ATC(m, cor_fun = WGCNA::bicor)
    )  

Value

A vector of numeric values with the same order as rows in the input matrix.

Author(s)

Zuguang Gu <z.gu@dkfz.de>

Examples

set.seed(12345)
nr1 = 100
mat1 = matrix(rnorm(100*nr1), nrow = nr1)

nr2 = 10
require(mvtnorm)
sigma = matrix(0.8, nrow = nr2, ncol = nr2); diag(sigma) = 1
mat2 = t(rmvnorm(100, mean = rep(0, nr2), sigma = sigma))

nr3 = 50
sigma = matrix(0.5, nrow = nr3, ncol = nr3); diag(sigma) = 1
mat3 = t(rmvnorm(100, mean = rep(0, nr3), sigma = sigma))

mat = rbind(mat1, mat2, mat3)
ATC_score = ATC(mat)
plot(ATC_score, pch = 16, col = c(rep(1, nr1), rep(2, nr2), rep(3, nr3)))

[Package cola version 1.2.1 Index]