associate: Cross Correlation Wrapper

In microbiome: Microbiome Analytics

Description

Cross-correlate columns of the input matrices.

Usage

associate(
  x,
  y = NULL,
  method = "spearman",
  p.adj.threshold = Inf,
  cth = NULL,
  order = FALSE,
  n.signif = 0,
  mode = "table",
  p.adj.method = "fdr",
  verbose = FALSE,
  filter.self.correlations = FALSE
)

Arguments

x	matrix (samples x features if annotation matrix)
y	matrix (samples x features if cross-correlated with annotations)
method	association method ('pearson', or 'spearman' for continuous; categorical for discrete)
p.adj.threshold	q-value threshold to include features
cth	correlation threshold to include features
order	order the results
n.signif	mininum number of significant correlations for each element
mode	Specify output format ('table' or 'matrix')
p.adj.method	p-value multiple testing correction method. One of the methods in p.adjust function ('BH' and others; see help(p.adjust)). Default: 'fdr'
verbose	verbose
filter.self.correlations	Filter out correlations between identical items.

Details

As the method=categorical (discrete) association measure for nominal (no order for levels) variables we use Goodman and Kruskal tau based on r-bloggers.com/measuring-associations-between-non-numeric-variables/

The p-values in the output table depend on the method. For the spearman and pearson correlation values, the p-values are provided by the default method in the cor.test function. For the categorical method, the p-value is estimated with the microbiome::gktau function (see the separate help page).

Value

List with cor, pval, pval.adjusted

Author(s)

Contact: Leo Lahti [e-mail]

References

See citation('microbiome')

Examples

data(peerj32)
d1 <- peerj32$microbes[1:20, 1:10]
d2 <- peerj32$lipids[1:20,1:10]
cc <- associate(d1, d2, method='pearson')

microbiome documentation built on Nov. 8, 2020, 5:08 p.m.