preprocessCSV: Pre-process scRNA-seq Count Matrix in CSV File

In ff98li/scRGNet: Perform scRNA-seq Analysis using Single-cell Graphical Neural Network (scGNN) Framework

Description

A function that reads a CSV file containing scRNA-seq counts and pre-process the scRNA-seq matrix with the given criteria. before feeding it to the scGNN framework for analysis.

Usage

preprocessCSV(
  path,
  log_transform = TRUE,
  cell_zero_ratio = 0.99,
  gene_zero_ratio = 0.99,
  geneSelectNum = 2000L,
  transpose = FALSE,
  toCSV = FALSE,
  outdir_path = file.path(getwd(), "output"),
  savename = "preprocessedCSV"
)

Arguments

path	A string the represents the path to the input file, where directories in the path are separated by "/", and the input data file should have extension .csv or .gz
log_transform	A Boolean value indicating whether to perform natural logarithmic transformation for every value of the data at the end of pre-processing or not. If TRUE, the log(x+1) of the original expression level will be calculated.
cell_zero_ratio	A double-precision value strictly greater than 0 and less than 1. It defines the maximum proportion of genes with zeros in cells. Default is 0.99, which means cells with more than 99% genes that are zeros will be filtered out.
gene_zero_ratio	A double-precision value strictly greater than 0 and less than 1. It defines the maximum proportion of zeros in genes. Default is 0.99, which means genes with more than 99% zero values will be filtered out.
geneSelectNum	A positive integer that defines how many most variant genes to keep. The default is 2000, which 2000 genes, ranked from most variant to least by their variances of expression values in each sample, will be retained in the final pre-processed data and be used for later analysis.
transpose	A Boolean value telling the function whether the input scRNA-seq matrix is given in a transposed manner, i.e. cells as rows and genes as columns. If TRUE, a transpose operation will be performed. Default is FALSE.
toCSV	A Boolean value indicating whether to save the pre-processed expression matrix to a CSV file. If TRUE a CSV file with <savename> will be saved to /output directory under user's current working directory.
outdir_path	A character vector that designates the path to the directory where the output CSV will be saved if toCSV is TRUE. This argument won't be used if toCSV is FALSE. There must be no path separator at the end. If the designated does not exist in the parent directory, it will create one automatically. Default is a directory named "output" under user's current R working directory.
savename	A character vector representing the name of the csv file that saves pre-processed scRNA-seq data. There is no strict rule for this. However, it is highly recommended to give it a meaningful name.

Value

Returns a scDataset with the pre-processed scRNA-seq expression values, where row names are cells and column names are genes.

References

\insertRef

scGNNscRGNet

Examples

# Example 2:
## Not run: 
# Accessing the demo gene_counts_small dataset available with the package
inputCountsPath <- system.file("extdata", "GSE138852_small.csv", package = "scRGNet")
# Preprocess the raw counts
counts <- preprocessCSV(path = inputCountsPath)

## End(Not run)
# Example 1:
## Not run: 
# Preprocess the whole scRNA-seq dataset
inputCountsPath <- system.file("extdata", "GSE138852_counts.csv.gz", package = "scRGNet")
counts <- preprocessCSV(path = inputCountsPath)
# Take ~20 seconds to process.

## End(Not run)

ff98li/scRGNet documentation built on Jan. 14, 2022, 4:58 a.m.