compute_SignatureFeatures-methods: Compute Signature Features Performs feature extraction for...

In hdsu-bioquant/ButchR: Non-Negative Matrix Factorization Suite

Compute Signature Features Performs feature extraction for all factorization ranks > 2. Returns an nmfExperiment object with computed signature features

Description

A K-means with k=2 is run over all rows of the W matrices. Resulting in a binary matrix for every factorization rank, with features in the rows and samples in the columns, in which 1 means that the features is contributing to the signature, and 0 it does not. The extraction of Signature Specific Features is not supported for k = 2

Usage

compute_SignatureFeatures(x)

## S4 method for signature 'ButchR_NMF'
compute_SignatureFeatures(x)

## S4 method for signature 'ButchR_integrativeNMF'
compute_SignatureFeatures(x)

## S4 method for signature 'ButchR_joinNMF'
compute_SignatureFeatures(x)

Arguments

x	An object of class ButchR_joinNMF.

Value

the same type of input object with computed signature features.

Examples

## Not run: 
data("leukemia")
nmf_exp <- run_NMF_tensor(leukemia$matrix, ranks = 3,
                             method = "NMF",
                             n_initializations = 2,
                             extract_features = FALSE)
nmf_exp <- compute_SignatureFeatures(nmf_exp)
SignatureSpecificFeatures(nmf_exp, k = 3)
SignatureSpecificFeatures(nmf_exp, k = 3, return_all_features = TRUE)

## End(Not run)
## Not run: 
data("leukemia")
nmf_exp <- run_NMF_tensor(leukemia$matrix, ranks = 3,
                             method = "NMF",
                             n_initializations = 2,
                             extract_features = FALSE)
nmf_exp <- compute_SignatureFeatures(nmf_exp)
SignatureSpecificFeatures(nmf_exp, k = 3)
SignatureSpecificFeatures(nmf_exp, k = 3, return_all_features = TRUE)

## End(Not run)
## Not run: 
norm_mat_list <- list(a = matrix(abs(rnorm(1000)), ncol = 10),
                      b = matrix(abs(rnorm(1000)), ncol = 10))
inmf_exp <- run_iNMF_tensor(norm_mat_list,
                            ranks = 2:5,
                            n_initializations     = 10,
                            iterations            = 10^4,
                            convergence_threshold = 40,
                            extract_features = FALSE)
inmf_exp <- compute_SignatureFeatures(inmf_exp)
SignatureSpecificFeatures(inmf_exp, k = 3)
SignatureSpecificFeatures(inmf_exp, k = 3, return_all_features = TRUE)

## End(Not run)
## Not run: 
norm_mat_list <- list(a = matrix(abs(rnorm(1000)), ncol = 10),
                      b = matrix(abs(rnorm(1000)), ncol = 10))
inmf_exp <- run_joinNMF_tensor(norm_mat_list,
                            ranks = 2:5,
                            n_initializations     = 10,
                            iterations            = 10^4,
                            convergence_threshold = 40,
                            extract_features = FALSE)
inmf_exp <- compute_SignatureFeatures(inmf_exp)
SignatureSpecificFeatures(inmf_exp, k = 3)
SignatureSpecificFeatures(inmf_exp, k = 3, return_all_features = TRUE)

## End(Not run)

hdsu-bioquant/ButchR documentation built on Jan. 28, 2023, 6:06 p.m.