all_calls: Run the method.

In angelesarzalluz/scfilters: A correlation-based method for quality filtering of single-cell RNAseq data

Description

The all_calls function calls all the functions that form the package pipeline, according to the provided parameters.

Usage

all_calls(data, import = TRUE, max_zeros = 0.75, randomizations = 25,
  top_method = "window_size", filter_parameter = 100,
  bin_method = "window_number", bin_parameter = 30,
  cor_method = "pearson")

Arguments

data	A file name, path or data frame containing the expression value table.
max_zeros	A number of type double. Indicates the maximum proportion of zeros allowed per row in the raw data.
randomizations	An integer. Number of times the top window will be randomized to generate a random negative control.
top_method	A string. Indicates the method for top window selection.
filter_parameter	An integer. Specifies a parameter for the top window selection method.
bin_method	A string. Indicates the method for binning of the genes.
bin_parameter	An integer. Specifies the parameter for the selected binning method.
cor_method	A string. Indicates the correlation method to use.
imort	A logical. Indicates whether raw data should be imported data from a file or path (TRUE) or is in a frame already loaded into R (FALSE).

Details

Arguments for this function are passed on to the each one of the functions in the package as indicated in this page, as they are called inside the function body.

randomizations is passed as an argument of the randomize_top function. Even though more random windows than specified by default (25) can be generated, it should be noted that bigger numbers may not make random negative control any more informative.

max_zeros is passed as an argument of the calculate_cvs function. For details on values and meaning, see this object's documentation.

When run, all_calls also displays useful messages that help the user track analysis progression, as well as information on the operations performed (i.e. number or size of bins generated, lowest mean expression value included in top window, etc.). If the retrieved values do not match what is expected, the pipeline can be stopped at any point and restarted from the beginning.

By default, the analysis is run creating the top window by size, selecting 100 genes, and the binning of the rest of the genes is done creating a specific number of windows (30 by default). For details on these methods and the parameters, see extract_top_genes and bin_scdata.

Important note: This function is used to run the analysis implemented in this package. Unless highly familiar with the method, it is not advised to run the different functions individually.

Value

A list containing all the results generated during the analysis. Each element on the list contains the results of a function call, and is named as follows:

• raw contains the raw data table as it was imported.

• cv contains the data frame output by calculate_cvs, which includes all the data plus the mean, cv and standard deviation columns.

• binned contains all the data as output by the combined call of extract_top_genes and bin_scdata, which includes the bin number column in the data frame.

• window_correlations contains the output of correlate, which outputs a list containing the vector of correlation coefficients of each window to the top window as a separate element.

• controls contains a nested list with the controls -as many as randomizations- computed for each window, as output by calling randomize_top and compute_control once per window.

• densities contains a list where the outputs of compute_density for each window are stored as a separate element.

• histogram contains a data frame with all the histogram values.

• info contains a string in which all the necessary information about the analysis is included. This should be used as the histogram title.

Note: use these names to retrieve information by subsetting using $.

angelesarzalluz/scfilters documentation built on May 10, 2019, 11:46 a.m.