fit_mp: Fit the Marchenko-Pastur distribution

In Siliegia/SIGMA: A clusterability measure for scRNA-seq data

Description

This function takes a data matrix as input and out puts the parameters asscociated to the MP distribution.

Usage

fit_mp(expr, sample = FALSE, cor = TRUE, nu = 50, p.val = 0.01)

Arguments

expr	a data matrix with cells in the columns and genes in the rows, preferably standardized gene-wise
sample	TRUE/FALSE, if the parameters should be estimated by random sampling or not, default is FALSE
cor	TRUE/FALSE, if the svd should be calculated on the correlation matrix or not (covariance matrix), default is TRUE
nu	the number of gene singular vectors to calculate in the process (the more, the more time expansive), default is 50
p.val	the p-value to be used in the test of normality for the singular vectors, default is 0.01

Value

A MP object:

• eigen: eigenvalues and vectors of the cell-cell correlation matrix

• maxEigen: maximum eigenvalue of the MP distribution

• minEigen: minimum eigenvalue of the MP distribution

• sig_vectors: singular vectors that lie significantly above the MP distribution

• M: the number of genes

• N: the number of cells

• svd: the singular value decomposition

• genes.used: genes that have been used for the calculation of the MP distribution

• p.value_mp_fit: p-value for the similarity of a MP distribution to the eigenvalues found as noise

• transcriptome_mode: the index of the transcriptome mode

• input_parameters: the inputs to the function

Examples

library(splatter)
data("splatO")
expr <- counts(splatO)
expr <- expr[rowSums(expr)>0,]

#Normalize and log-transform the data
expr.norm <- t(t(expr)/colSums(expr))*10000
expr.norm.log <- log(expr.norm + 1)

expr.scale <- t(scale(t(expr.norm.log)))
L <- fit_mp(expr.scale)

Siliegia/SIGMA documentation built on Dec. 18, 2021, 2 p.m.