remove_signs_redundant: Remove redundant signs using semantic similarity matrices.

In keita-iida/ASURAT: Functional annotation-driven unsupervised clustering for single-cell data

Remove redundant signs using semantic similarity matrices.

Description

This function removes redundant signs using semantic similarity matrices.

Usage

remove_signs_redundant(
  sce = NULL,
  similarity_matrix = NULL,
  threshold = NULL,
  keep_rareID = NULL
)

Arguments

sce	A SingleCellExperiment object.
similarity_matrix	A semantic similarity matrix.
threshold	A threshold value of semantic similarity, used for regarding biological terms as similar ones
keep_rareID	If TRUE, biological terms with the larger ICs are kept.

Value

A SingleCellExperiment object.

Examples

data(pbmc_eg)
data(human_GO_eg)
mat <- t(as.matrix(SummarizedExperiment::assay(pbmc_eg, "centered")))
pbmc_cormat <- cor(mat, method = "spearman")
pbmcs <- list(GO = pbmc_eg)
S4Vectors::metadata(pbmcs$GO) <- list(sign = human_GO_eg[["BP"]])
pbmcs$GO <- remove_signs(sce = pbmcs$GO, min_ngenes = 2, max_ngenes = 1000)
pbmcs$GO <- cluster_genesets(sce = pbmcs$GO, cormat = pbmc_cormat,
                             th_posi = 0.24, th_nega = -0.20)
pbmcs$GO <- create_signs(sce = pbmcs$GO, min_cnt_strg = 2, min_cnt_vari = 2)
pbmcs$GO <- remove_signs_redundant(
  sce = pbmcs$GO, similarity_matrix = human_GO_eg$similarity_matrix$BP,
  threshold = 0.80, keep_rareID = TRUE)
# The results are stored in `metadata(pbmcs$GO)$sign_SCG`,
# `metadata(pbmcs$GO)$sign_VCG`, `metadata(pbmcs$GO)$sign_all`,
# and if there exist, `metadata(pbmcs$GO)$sign_SCG_redundant` and
# `metadata(pbmcs$GO)$sign_VCG_redundant`.

keita-iida/ASURAT documentation built on Nov. 23, 2022, 6:50 p.m.