MIXTURE: MIXTURE

In elmerfer/MIXTURE: MIXTURE: a nu-SVR based recursive feature extraction mdoel with noise constraints for molecular signature deconvolution of bulk tumor tissues

MIXTURE

Description

the MIXTURE algorithm solves Y = X*B estimating the B coefficients or cell type proportions (B>0)

Usage

MIXTURE(
  expressionMatrix,
  signatureMatrix,
  iter = 100,
  functionMixture = nu.svm.robust.RFE,
  useCores = 1L,
  verbose = FALSE,
  nullDist = c("none", "SingleSubjectTest", "PopulationBased"),
  fileSave
)

Arguments

expressionMatrix	a GxS gene expression matrix. with row names genes ID as in the signature Matrix. Genes in rows, samples in columns
signatureMatrix	a NxL gene signature matrix for L cell types
iter	integer, number of iterations to estimate the p values
functionMixture	: (default nu.svm.robust.RFE) the deconvolution function: cibersort, nu.svm.robust.RFE, ls.rfe.abbas (typed without quotation marks)
useCores	: (integer) the number of cpus to use.
verbose	: (logical) if TRUE, msgs on the screen
nullDist	: (character) select the null distribution mode for p.valu estimation of the regression estimate one of the followings : "none" , "PopulationBased" use the whole expressionMatrix to draw "iter" random samples at once. Then all the models are compared against this null distribution
filesave	: (character) the name of the output EXCEL(r) file in xlsx format

Value

an S3 object of class MIXTURE (list) with the following slots: Subjects : a list with the following slots: MIXabs : a data.frame with S rows, and L columns with the absoluite regression coefficients MIXprop : a data.frame with S rows, and L columns with the proportions ACCmetric : a data.frame with S rows, and RMSEa (Root mean squared error from absolute coeficientes), RMSEp (RMSE with proportions coefficientes) Ra (correlation absolute coeffs) Rp (correlation proportion coeffs) BestParams (Best model parameters)Iter (number of itereation of the RFE to reach the best model)

PermutedMatrix : the simulated wxpression matrix for null distribution estimation (if nullDist == "none", PermutedMatrix = NULL) p.values : The p.values for each subject and for each of the ACC metrix columns (if nullDist == "none", p.value = NULL) method : the null distribution method (nullDist) usedGenes : the intersection gene list between expressionMatrix and signatureMatrix

elmerfer/MIXTURE documentation built on Aug. 20, 2024, 8:03 p.m.