FUCHIKOMA: Revealing differentially expressed genes using nonlinear...

In rikenbit/FUCHIKOMA: Revealing differentially expressed genes using nonlinear dimensionality reduction for single cell RNA-Seq data

Description

This package was not yet installed at build time.

Usage

fuchikoma(data, cores=NULL, mode = c("Supervised", "Unsupervised", "Mix", "tSNE"), weight=c(0.5,0.5), Comp = NULL, label = FALSE, cat.type = c("simple", "one_vs_rest", "each", "two"), n.eigs = 10, algorithm = c("song", "brute"), per.rej = 10, threshold = 0.01, verbose=FALSE, dropout=10, sigma=15, perplexity = 30)

Arguments

data	A data matrix, in which the row means genes and column means cells or sample.
cores	The number of using mathine cores. (default: automaticaly detected number by doParallel package)
mode	When Supervised is specified, fuchikoma uses label paramter. When Unsupervised is specified, fuchikoma uses Comp parameter for specifying which diffusion components should be used. When Mixed is specified, fuchikoma uses both mode (default:Supervised)
weight	Weight for integrating two gram matrix, when mode is specified as Mix (default:0.5 vs 0.5)
Comp	When mode is specified as Unsupervised, Comp must be specified such as c(1,2). (default:FALSE)
label	When mode is specified as Supervised, label must be specified such as c(1,1,1,2,2,2,3,3) (default:FALSE)
cat.type	Type of categorical kernel of CatKernel (default: simple)
n.eigs	Number of eigenvectors/values to return (default: 20)
algorithm	brute means single gene rejection strategies and song means fixed percent of genes rejection strategies in each iteration step of fuchikoma. (default:song)
per.rej	When algorithm is specified as song, per.rej must be specified such as 20 (default:10)
threshold	In each iteration step, if the difference of HSIC in the step and max value of previous HSICs is lower than threshold, iteration will be halted (default: 0.01).
verbose	verbose option (default: FALSE).
dropout	Threshold when the remaining gene is few (default: 10).
sigma	Parameter used in DiffusionMap of destiny (default: 15).
perplexity	Parameter of tSNE (default: 30).

Value

DEGs.HSICs	Differentially expressed genes (DEGs) and HSIC values
DEGs.Pvals	Pvalues of DEGs
All.HSICs	HSICs value of all genes
All.Pvals	Pvalues of all genes

Author(s)

Koki Tsuyuzaki, Haruka Ozaki, Mika Yoshimura, Itoshi Nikaido

Maintainer: Koki Tsuyuzaki <[e-mail]>

References

Laleh Haghverdi et al. (2015) Diffusion maps for high-dimensional single-cell analysis of differentiation data. Bioinformatics, 31(18), 2989-2998

Le Song et al. (2007) Gene selection via the BAHSIC family of algorithms, Bioinformatics, 23(13), i490-i498

Y-h Taguchi et al. (2015) Principal component analysis-based unsupervised feature extraction applied to in silico drug discovery for posttraumatic stress disorder-mediated heart disease, BMC Bioinformatics, 16(139)

Aaditya Ramdas et al. (2015) Nonparametric Independence Testing for Small Sample Sizes, IJCAI-15

See Also

DiffusionMap

Examples

# three cell data
CellA <- data.frame(matrix(rnorm(50*20), nrow=50, ncol=20))
CellB <- data.frame(matrix(rnorm(50*20), nrow=50, ncol=20))

# DEGs definition
CellA[1:10, ] <- CellA[1:10, ] + 10 * matrix(runif(10*20), nrow=10, ncol=20)
CellB[11:20, ] <- CellB[1:10, ] + 10 * matrix(runif(10*20), nrow=10, ncol=20)
# testdata
testdata2 <- data.frame(CellA, CellB)
colnames(testdata2) <- c(paste0("CellA_", 1:20), paste0("CellB_", 1:20))
rownames(testdata2) <- paste0("Gene", 1:nrow(testdata2))

# label
label2 <- c(rep(1, 20), rep(2, 20))

res <- fuchikoma(data=testdata2, cores=1, label=label2)

rikenbit/FUCHIKOMA documentation built on May 27, 2019, 9:09 a.m.