computeCorrelation: Compute weighted correlation network
In kwanjeeraw/metabox: metabox

Description Usage Arguments Details Value Note Author(s) References See Also Examples

compute the correlation network of entities from one or two quantified data sets, see details.

1	computeCorrelation(x, y, xtype, ytype, internalid, coef, pval, method, returnas)

`x`	a data frame of raw or quantified data e.g. gene expression data, metabolite intensities. Columns are samples and rows are molecular entities e.g. genes, proteins or compounds.
`y`	a data frame of raw or quantified data e.g. gene expression data, metabolite intensities. Columns are samples and rows are molecular entities e.g. genes, proteins or compounds. See details.
`xtype`	a string specifying a node type (default = NULL). If provided and the database is installed, node attributes will be automatically retrieved from the database. For database query, the value can be one of compound, protein, gene, rna, dna.
`ytype`	a string specifying a node type (default = NULL).
`internalid`	a logical value indicating whether the network nodes are neo4j ids, if TRUE (default). See `convertId` for how to convert ids. It has no effect if `xtype` = NULL or there is no database installed.
`coef`	a numeric value specifying the minimum absolute correlation coefficient to be included in the output (from 0 to 1, default is 0.7).
`pval`	a numeric value specifying the maximum p-value to be included in the output (default is 0.05).
`method`	a string specifying method for computing correlation. It can be one of pearson, kendall, spearman (default). See `cor` for details.
`returnas`	a string specifying output type. It can be one of dataframe, list, json. Default is dataframe.

The function wraps around corAndPvalue function of WGCNA. Correlation coefficients, p-values and correlation directions are calculated. The correlation coefficients are continuous values between -1 (negative correlation) and 1 (positive correlation), with numbers close to 1 or -1, meaning very closely correlated.

x and y are data frame in which columns are samples and rows are entities. If y is given, then the correlations between the x entities and y entities are computed. Otherwise the correlations between x entities are computed.

list of network information with the following components:

nodes:

id = node id or node neo4j id

gid = node id or node grinn id

nodename = node id or node name

nodelabel = node type if provided

edges:

source, target = node id or node neo4j id

coef = correlation coefficient

pval = p-value

direction = correlation direction

type = relationship type

Return empty list if error or found nothing.

If only one data set x is provided and the database is installed, node attributes will be automatically retrieved from the database. Otherwise node attributes will be the original input.

Kwanjeera W kwanich@ucdavis.edu

Langfelder P. and Horvath S. (2008) WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics, 9:559

Dudoit S., Yang YH., Callow MJ. and Speed TP. (2002) Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments, STATISTICA SINICA, 12:111

Langfelder P. and Horvath S. Tutorials for the WGCNA package http://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/Rpackages/WGCNA/Tutorials/index.html

cor, corAndPvalue

# Compute a correlation network from compound data
#dt <- data.frame(id=seq(32), mtcars, row.names = NULL) #data frame of x
#result <- computeCorrelation(x = dt, xtype = "compound", coef=0.9, pval = 1e-10)
# Compute a correlation network from two data sets
#dtX <- data.frame(id=seq(16), mtcars[1:16,], row.names = NULL) #data frame of x
#dtY <- data.frame(id=seq(17,32), mtcars[17:32,], row.names = NULL) #data frame of x
#result <- computeCorrelation(x=dtX, y=dtY, coef=0.9, pval = 1e-7)