Skip to contents

Annotate NMF factors with Bagaev et al. (2021) MFP subtypes

map_factors_to_TME.Rd

For a single cancer type, matches TCGA patients present in the NMF factor score matrix Z to the Bagaev annotation, then computes Spearman correlations between each factor and each one-hot-encoded MFP subtype (IE, IE/F, F, D). The best-matching subtype per factor is returned; factors with a maximum absolute correlation below 0.15 are labelled "uncharacterized".

Usage

map_factors_to_TME(cancer_name, Z, plot = TRUE, file_name = NULL)

Arguments

cancer_name

Character. Cancer type abbreviation matching the TCGA_project column of the internal annotation (case-insensitive, e.g. "skcm").

Z

Numeric matrix. Samples x factors NMF score matrix (row names = TCGA barcodes).

plot

Logical. If TRUE (default), saves a boxplot of factor scores by MFP group.

file_name

Optional character. File path prefix for saving output plots.

Value

A data frame with columns factor, best_MFP, kw_pval, median_IE, median_IEF, median_F, median_D, n_samples, or NULL if fewer than 10 patients are matched.