individual_clustering: individual_clustering

In yycunc/SAMEclustering: SAME: Single-cell RNA-seq Aggregated clustering via Mixture model Ensemble for Single-cell RNA-seq Data

Description

This function performs single-cell clustering using five state-of-the-art methods, SC3, CIDR, Seurat, tSNE+kmeans and SIMLR.

Usage

individual_clustering(inputTags, mt_filter = TRUE, mt.pattern = "^MT-", mt.cutoff = 0.1, 
percent_dropout = 10, SC3 = TRUE, gene_filter = FALSE, svm_num_cells = 5000, CIDR = TRUE, nPC.cidr = NULL, 
Seurat = TRUE, nGene_filter = TRUE, low.genes = 200, high.genes = 8000, nPC.seurat = NULL, resolution = 0.7, 
tSNE = TRUE, dimensions = 3, perplexity = 30, tsne_min_cells = 200, tsne_min_perplexity = 10, var_genes = NULL, 
SIMLR = TRUE, diverse = TRUE, SEED = 1)

Arguments

inputTags	a G*N matrix with G genes and N cells.
mt_filter	is a boolean variable that defines whether to filter outlier cells according to mitochondrial gene percentage. Default is "TRUE".
mt.pattern	defines the pattern of mitochondrial gene names in the data, for example, mt.pattern = "^MT-" for human and mt.pattern = "^mt-" for mouse. Only applied when mt_filter = TRUE. Default is mt.pattern = "^MT-".
mt.cutoff	defines a high cutoff of mitochondrial percentage (Default is 0.1) that cells having higher percentage of mitochondrial gene are filtered out, when mt_filter = TRUE
percent_dropout	defines a low cutoff of gene percentage that genes expressed in less than percent_dropout% or more than (100 - percent_dropout)% of cells are removed for CIDR, tSNE + k-means, and SIMLR clustering. Default is 10.
SC3	a boolean variable that defines whether to cluster cells using SC3 method. Default is "TRUE".
gene_filter	is a boolean variable that defines whether to perform gene filtering before SC3 clustering, when SC3 = TRUE.
svm_num_cells	if SC3 = TRUE, then defines the mimimum number of cells above which SVM will be run.
CIDR	is a boolean parameter that defines whether to cluster cells using CIDR method. Default is "TRUE".
nPC.cidr	defines the number of principal coordinates used in CIDR clustering, when CIDR = TRUE. Default value is esimated by nPC of CIDR.
Seurat	is a boolean variable that defines whether to cluster cells using Seurat method. Default is "TRUE".
nGene_filter	is a boolean variable that defines whether to filter outlier cells according to unique gene count before Seurat clustering. Default is "TRUE".
low.genes	defines a low cutoff of unique gene counts (Default is 200) that cells expressing less than 200 genes are filtered out, when nGene_filter = TRUE.
high.genes	defines a high cutoff of unique gene counts (Default is 8000) that cells expressing more than 8000 genes are filtered out, when nGene_filter = TRUE.
nPC.seurat	defines the number of principal components used in Seurat clustering, when Seurat = TRUE. Default is nPC.seurat = nPC.cidr.
resolution	defines the value of resolution used in Seurat clustering, when Seurat = TRUE. Default is resolution = 0.7.
tSNE	is a boolean variable that defines whether to cluster cells using t-SNE + k-means method. Default is "TRUE".
dimensions	sets the number of dimensions wanted to be retained in t-SNE step. Default is 3.
perplexity	sets the perplexity parameter for t-SNE dimension reduction. Default is 30 when number of cells >=200.
tsne_min_cells	defines the number of cells in input dataset below which tsne_min_perplexity=10 would be employed for t-SNE step. Default is 200.
tsne_min_perplexity	sets the perplexity parameter of t-SNE step for small datasets (number of cells <200).
var_genes	defines the number of variable genes used by t-SNE analysis, when tSNE = TRUE.
SIMLR	is a boolean variable that defines whether to cluster cells using t-SNE + k-means method. Default is "TRUE".
diverse	is a boolean parameter that defines whether to take a subset of 4 out of 5 most diverse sets of clustering. Can only be implemented when all 5 booleans for individual methods are set to TRUE. Default is "TRUE".
SEED	sets the seed of the random number generator. Setting the seed to a fixed value can produce reproducible clustering results.

Value

a matrix of indiviudal clustering results is output, where each row represents the cluster results of each method.

Author(s)

Ruth Huh <[e-mail]>, Yuchen Yang <[e-mail]>, Yun Li <[e-mail]>

References

Huh, R., Yang, Y., Jiang, Y., Shen, Y., Li, Y. (2020) SAME-clustering: Single-cell Aggregated Clustering via Mixture Model Ensemble. Nucleic Acid Research, 48: 86-95

Examples

# Load the example data data_SAME
data("data_SAME")

# Zheng dataset
# Run individual_clustering
cluster.result <- individual_clustering(inputTags=data_SAME$Zheng.expr, SEED=123)

yycunc/SAMEclustering documentation built on May 6, 2021, 6:05 p.m.