BinaryPCA: Performs Binary PCA (as outlined in our paper). This function...

In scBFA: A dimensionality reduction tool using gene detection pattern to mitigate noisy expression profile of scRNA-seq

Description

Performs Binary PCA (as outlined in our paper). This function take the input of gene expression profile and perform PCA on gene detection pattern

Usage

BinaryPCA(scData, X = NULL, scale. = FALSE, center = TRUE)

Arguments

scData	can be a raw count matrix, in which rows are genes and columns are cells; can be a seurat object; can be a SingleCellExperiment object.
X	N by C covariate matrix,e.g batch effect, in which rows are cells,columns are number of covariates. If no such covariates available X = NULL
scale.	Logical value isndicating whether the variables should be scaled to have unit variance before the analysis takes place. In general scaling is not advisable, since we think the variance in the gene detection space is potentially associated with celltypes (e.g cell type specific markers)
center	Logical value indicating whether the variables should be shifted to be zero centered

Value

A list with class "prcomp",containing the following components:

sdev: the standard deviations of the principal components (i.e., the square roots of the eigenvalues of the covariance/correlation matrix, though the calculation is actually done with the singular values of the data matrix).

rotation: the matrix of variable loadings (i.e., a matrix whose columns contain the eigenvectors). The function princomp returns this in the element loadings.

x: the rotated data (the centred (and scaled if requested) data multiplied by the rotation matrix) is returned. Hence, cov(x) is the diagonal matrix diag(sdev^2).

center, scale. centering and scaling used, or FALSE.

Examples

## Working with Seurat or SingleCellExperiment object

library(Seurat)
library(SingleCellExperiment)

## Input expression profile, 5 genes x 3 cells

GeneExpr = matrix(rpois(15,1),nrow = 5,ncol = 3)
rownames(GeneExpr) = paste0("gene",seq_len(nrow(GeneExpr)))
colnames(GeneExpr) = paste0("cell",seq_len(ncol(GeneExpr)))
celltype = as.factor(sample(c(1,2,3),3,replace = TRUE))

## Create cell level technical batches

batch = sample(c("replicate 1","replicate 2","replicate 2"))
X = matrix(NA,nrow = length(batch),ncol = 1)
X[which(batch =="replicate 1"), ] = 0
X[which(batch =="replicate 2"), ] = 1
rownames(X) = colnames(GeneExpr)

##run BFA with raw count matrix

bpca_model = BinaryPCA(scData = GeneExpr,X = scale(X))

## Create Seurat object for input to BFA

scData = CreateSeuratObject(counts = GeneExpr,project = "sc",min.cells = 0)

## Standardize the covariate matrix should be a default operation

bpca_model = BinaryPCA(scData = scData, X = scale(X))

## Build the SingleCellExperiment object for input to BFA

## Set up SingleCellExperiment class

sce <- SingleCellExperiment(assay = list(counts = GeneExpr))

## Standardize the covariate matrix should be a default operation

bpca_model = BinaryPCA(scData = sce, X = scale(X))

scBFA documentation built on Nov. 8, 2020, 5:24 p.m.