DSAproportions. CellMix 1.6

Description

DSAproportions implements the method proposed by Zhong et al. (2013) as part of the DSA algorithm. This method estimates cell proportions from mixed sample expression data, given a set of markers, i.e. features that are known to be exclusively expressed by a single cell type in the mixture.

.DSAproportions is the internal -- non-exported -- function that actually performs the estimation. It expects a list of valid indexes for each cell type whose proportions are to be estimated.

Usage

DSAproportions(x, markers, log = NULL, match.names = TRUE, verbose = FALSE)

.DSAproportions(x, markers)

Arguments

x: target mixed expression matrix
markers: marker specification
log: indicates if the data are in log-scale or should be converted to linear-scale. This is relevant because the DSA algorithm assumes that the input mixed data are in linear scale (i.e. not log-trasnformed). If NULL, then data's scale is detected by link{is_logscale} and conversion to linear-scale is performed if necessary. If TRUE the data is exponentialised (using log base 2). If FALSE the data is left unchanged (the detected log scale is displayed in verbose mode). If a number, then it is used as the log base to exponentialise the data.
match.names: logical that indicates if marker names should be first matched against the data, to convert them into integer indexes. If codeFALSE, then markers should be a list of integer vectors that corresponds to the indexes of markers for each cell types.
verbose: verbose level

Value

a matrix of proportions with the same number of columns as x and as many rows as elements in markers (i.e. cell types).

References

Zhong Y, Wan Y, Pang K, Chow LM and Liu Z (2013). "Digital sorting of complex tissues for cell type-specific gene expression profiles." _BMC Bioinformatics_, *14*(1), pp. 89. ISSN 1471-2105, , .

Examples


# random data
x <- rmix(3, 100, 20)
m <- getMarkers(x)

# estimate proportions
p <- DSAproportions(x, m)
# plot against true proportions
profplot(x, p)