DCARS: DCARS
In shazanfar/DCARS: Differential Correlation across Ranked Samples

Description Usage Arguments Value Examples

The DCARS function performs testing for differential correlation across ranked samples.

DCARS(dat, xname, yname, W = NULL, rangeMin = 0, wcormin = 0,
  statmin = 0, weightedConcordanceFunction = weightedPearson,
  weightedConcordanceFunctionW = "vector",
  extractTestStatisticOnly = FALSE, extractWcorSequenceOnly = FALSE,
  plot = FALSE, niter = 100,
  extractPermutationTestStatistics = FALSE, verbose = FALSE,
  forceBlock = FALSE, WcorSequenceSummaryFun = sd, ...)

`dat`	a genes x samples gene expression rank matrix, should be already converted to ranks with first column lowest survival and last column highest survival
`xname`	name of row of dat to test together with yname
`yname`	name of row of dat to test together with xname
`W`	weight matrix for weighted correlations,
`rangeMin`	minimum range of weighted correlation vector to include for permutation testing
`wcormin`	minimum absolute value weighted correlation vector to include for permutation testing
`statmin`	minimum value DCARS test statistic to include for permutation testing
`weightedConcordanceFunction`	concordance function to use, defaults to weighted Pearson correlation. User can provide their own function, with arguments fun(x,y,w).
`weightedConcordanceFunctionW`	either "vector" or "matrix", determining if the function given in weightedConcordanceFunction argument takes in a single vector as input for w, or a matrix.
`extractTestStatisticOnly`	if TRUE, extract only the DCARS test statistic without permutation testing
`plot`	if TRUE plot observed weighted correlatin vector
`niter`	number of iterations for permutation testing
`extractPermutationTestStatistics`	if TRUE, extract only the DCARS test statistic without permutation testing
`verbose`	if TRUE, print updates
`forceBlock`	if TRUE, this converts all positive values in weight matrix W to 1, and calculates weighted correlation on the resulting subset. Default FALSE
`WcorSequenceSummaryFun`	Function that defines how the local weighted correlations are summarised, default is the function sd for standard deviation.
`...`	additional arguments passing on to weightMatrix()
`extractWcorSequence`	if TRUE, extract only the weighted correlation vector without permutation testing

either single value (p-value or test statistic), vector (local weighted correlation), or list (combination of above) depending on the input parameters.

data(STRING)
data(SKCM)
SKCM_rank = t(apply(SKCM,1,rank))

# highly significantly DCARS gene pair: SKP1 and SKP2
# calculates p-value based on permutation
DCARS(SKCM_rank,"SKP1","SKP2",plot=TRUE)
# extract only the test statistic
DCARS(SKCM_rank,"SKP1","SKP2", extractTestStatisticOnly = TRUE)

# not significantly DCARS gene pair: EIF3C and EIF5B
# calculates p-value based on permutation
DCARS(SKCM_rank,"EIF3C","EIF5B",plot=TRUE)
# extract only the test statistic
DCARS(SKCM_rank,"EIF3C","EIF5B", extractTestStatisticOnly = TRUE)

# build weight matrix
W = weightMatrix(ncol(SKCM_rank), type = "triangular", span = 0.5, plot = TRUE)

# extract DCARS test statistics
SKCM_stats = DCARSacrossNetwork(SKCM_rank,edgelist = STRING,
                                W = W, extractTestStatisticOnly = TRUE,
                                verbose = FALSE)
sort(SKCM_stats,decreasing=TRUE)[1:10]