R/pairwiseLinearFit.R

In ctlab/ClusDec: Clustering approach to solve complete Deconvolution problem

#' Pairwise linear fits between genes
#'
#' Performs pairwise linear fit gene1 ~ k * gene2
#'
#' @param X matrix, given gene expression dataset, rows supposed to be genes
#'
#' @return list containing two N x N matrices: R2 and slopes
#' @export
pairwiseLinearFit <- function(X) {
    results <- .Call('clusdec_pairwiseLinearFit', t(X), PACKAGE="clusdec")
    rownames(results$slopes) <- rownames(X)
    colnames(results$slopes) <- rownames(X)
    rownames(results$R2) <- rownames(X)
    colnames(results$R2) <- rownames(X)
    return(results)
}


#' Pairwise Deming regression fits between genes
#'
#' Performs pairwise linear fit gene1 ~ k * gene2
#'
#' @param X matrix, given gene expression dataset, rows supposed to be genes
#'
#' @return list containing two N x N matrices: R2 and slopes
#' @export
pairwiseDemingRegression <- function(X) {
    results <- .Call('clusdec_pairwiseDemingRegression', t(X), PACKAGE="clusdec")
    rownames(results$slopes) <- rownames(X)
    colnames(results$slopes) <- rownames(X)
    rownames(results$R2) <- rownames(X)
    colnames(results$R2) <- rownames(X)
    return(results)
}

#' Deconvolution by it self
#'
#' @return list containing basis, proportions and reconstructed results
#' @export
deconvolve <- function(mixedData, gg, coefMatrix, coef, dims) {
    results <- .Call('clusdec_deconvolve', mixedData, gg, coefMatrix, coef, dims, PACKAGE="clusdec")
    rownames(results$results) <- rownames(mixedData)
    colnames(results$results) <- colnames(mixedData)
    rownames(results$basis) <- rownames(mixedData)
    colnames(results$basis) <- paste0("Cell type ", 1:dims)
    rownames(results$proportions) <- paste0("Cell type ", 1:dims)
    colnames(results$proportions) <- colnames(mixedData)
    return(results)
}