imputeByAttractors: Imputation based on co-expressed gene attractor signatures

In zky0708/2DImpute: 2DImpute: Imputing scRNA-seq data from correlations in both dimensions

Description

This function imputes dropout-suspected zeros by borrowing information from co-expressed genes in the identified co-expressed attractor signatures.

Usage

imputeByAttractors(
  exprs,
  dropout_ind,
  attractor_list,
  mi_threshold = 0.4,
  verbose = TRUE
)

Arguments

exprs	A log-transformed expression matrix, where rows and columns represent genes and cells, respectively.
dropout_ind	A matrix where each row contains the indices of zero entries (first column for gene name, second column for cell ID) that are suspected as dropouts.
attractor_list	A list where each element contains the identified co-expressed signature.
mi_threshold	MI threshold (between 0 to 1) to determine the gene membership of signatures. Genes with greater than or equal to the threshold are considered to be associated with the signature. Default is 0.2.

Details

For each gene of interest, we use the top 20 genes involved in the same signature as the predictive variables to fit a linear regression model using cells where all those genes are expressed.

Value

A list consisting of the following components:

imputed	The imputed version of the expression matrix
dropout_ind	An updated indices matrix of dropout-suspected zero entries, where in each contains the indices of the remaining dropouts after this step of imputation.

zky0708/2DImpute documentation built on March 12, 2020, 3:13 a.m.