Keep Only Significant Values in a Symmetric Matrix

Replaces entries in a symmetric matrix (typically a correlation matrix) with NA wherever the corresponding p-value exceeds a significance threshold - retaining only the statistically supported associations. Designed as a pre-processing step for pharos::plotWeb() and pharos::plotCor().

Usage

keepSig(
  m,
  p = NULL,
  data = NULL,
  sig.level = 0.05,
  method = c("pearson", "spearman", "kendall"),
  diag = TRUE
)

Arguments

m

a symmetric numeric matrix. Typically the output of stats::cor(), but any symmetric matrix of effect sizes or association measures is accepted.

p

a matrix of p-values of the same dimension as m. If NULL (default), p-values are computed from pairwise correlation tests on the columns of data (requires data).

data

an optional numeric data frame or matrix. Used to compute p when p = NULL. Ignored if p is supplied.

sig.level

numeric; significance threshold. Entries where p > sig.level are replaced by NA. Default 0.05.

method

character; the correlation test method passed to stats::cor.test() when computing p-values from data. One of "pearson" (default), "spearman", or "kendall".

diag

logical; if TRUE (default), the diagonal is kept as-is (typically 1 for correlation matrices). If FALSE, the diagonal is also set to NA.

Value

matrix with the same dimensions and dimnames as m, with NA wherever p > sig.level

See also

pharos::plotWeb, pharos::plotCor, lumen::corTest, stats::cor.test

Other assoc.continuous: corPart(), corPolychor(), findCorrX(), hoeffdingD(), pearsonCor(), spearmanCor()

Examples

# compute p-values on the fly from the raw data
keepSig(cor(mtcars), data = mtcars)
#>             mpg        cyl       disp         hp       drat         wt
#> mpg   1.0000000 -0.8521620 -0.8475514 -0.7761684  0.6811719 -0.8676594
#> cyl  -0.8521620  1.0000000  0.9020329  0.8324475 -0.6999381  0.7824958
#> disp -0.8475514  0.9020329  1.0000000  0.7909486 -0.7102139  0.8879799
#> hp   -0.7761684  0.8324475  0.7909486  1.0000000 -0.4487591  0.6587479
#> drat  0.6811719 -0.6999381 -0.7102139 -0.4487591  1.0000000 -0.7124406
#> wt   -0.8676594  0.7824958  0.8879799  0.6587479 -0.7124406  1.0000000
#> qsec  0.4186840 -0.5912421 -0.4336979 -0.7082234         NA         NA
#> vs    0.6640389 -0.8108118 -0.7104159 -0.7230967  0.4402785 -0.5549157
#> am    0.5998324 -0.5226070 -0.5912270         NA  0.7127111 -0.6924953
#> gear  0.4802848 -0.4926866 -0.5555692         NA  0.6996101 -0.5832870
#> carb -0.5509251  0.5269883  0.3949769  0.7498125         NA  0.4276059
#>            qsec         vs         am       gear       carb
#> mpg   0.4186840  0.6640389  0.5998324  0.4802848 -0.5509251
#> cyl  -0.5912421 -0.8108118 -0.5226070 -0.4926866  0.5269883
#> disp -0.4336979 -0.7104159 -0.5912270 -0.5555692  0.3949769
#> hp   -0.7082234 -0.7230967         NA         NA  0.7498125
#> drat         NA  0.4402785  0.7127111  0.6996101         NA
#> wt           NA -0.5549157 -0.6924953 -0.5832870  0.4276059
#> qsec  1.0000000  0.7445354         NA         NA -0.6562492
#> vs    0.7445354  1.0000000         NA         NA -0.5696071
#> am           NA         NA  1.0000000  0.7940588         NA
#> gear         NA         NA  0.7940588  1.0000000         NA
#> carb -0.6562492 -0.5696071         NA         NA  1.0000000

# stricter threshold, and drop the diagonal as well
keepSig(cor(swiss), data = swiss, sig.level = 0.01, diag = FALSE)
#>                   Fertility Agriculture Examination  Education   Catholic
#> Fertility                NA          NA  -0.6458827 -0.6637889  0.4636847
#> Agriculture              NA          NA  -0.6865422 -0.6395225  0.4010951
#> Examination      -0.6458827  -0.6865422          NA  0.6984153 -0.5727418
#> Education        -0.6637889  -0.6395225   0.6984153         NA         NA
#> Catholic          0.4636847   0.4010951  -0.5727418         NA         NA
#> Infant.Mortality  0.4165560          NA          NA         NA         NA
#>                  Infant.Mortality
#> Fertility                0.416556
#> Agriculture                    NA
#> Examination                    NA
#> Education                      NA
#> Catholic                       NA
#> Infant.Mortality               NA

# the intended use is as a pre-processing step for the plots
pharos::plotWeb(keepSig(cor(mtcars), data = mtcars))


# supply a pre-computed p-value matrix
m <- cor(mtcars)
p <- outer(
  (vars <- colnames(mtcars)), vars,
  Vectorize(function(v1, v2)
    cor.test(mtcars[[v1]], mtcars[[v2]])$p.value)
)
dimnames(p) <- list(vars, vars)
plotWeb(keepSig(m, p = p))