GI_screen: Perform genetic interaction screen

In ytakemon/GINIR: Genetic inteRaction and EssenTiality neTwork mApper

Perform genetic interaction screen

Description

Compares dependency probabilities of mutant and control groups to determine whether the mutant group are resistant or sensitive to specific gene perturbations.

Usage

GI_screen(
  control_id = NULL,
  mutant_id = NULL,
  gene_list = NULL,
  core_num = NULL,
  output_dir = NULL,
  data_dir = NULL,
  filename = NULL,
  test = FALSE
)

Arguments

control_id	string, A vector containing two or more DepMap_id, Default: NULL
mutant_id	string, A vector containing two or more DepMap_id, Default: NULL
gene_list	string, A vector containing a list of Hugo symbols to subset the screen and to perform a small in-silico screen, Default: NULL
core_num	integer, Number of cores to run analysis, Default: NULL
output_dir	string, Full path to where output file should be saved, Default: NULL
data_dir	string Path to GRETTA_data
filename	string name of file without the '.csv' extension.
test	logical, For test_that to shorten computational time for testing

Details

Description of output data frame

• GeneName_ID - Hugo symbol with NCBI gene ID

• GeneNames - Hugo symbol

• ⁠_median⁠, ⁠_mean⁠, ⁠_sd⁠, ⁠_iqr⁠ - Control and mutant group's median, mean, standard deviation (sd), and interquartile range (iqr) of dependency probabilities. Dependency probabilities range from zero to one, where one indicates a essential gene (ie. KO of gene was lethal) and zero indicates a non-essential gene (KO of gene was not lethal)

• Pval - P-value from Mann Whitney U test between control and mutant groups.

• Adj_pval - BH-adjusted P-value.

• log2FC_by_median - Log2 normalized median fold change of dependency probabilities (mutant / control).

• log2FC_by_mean - Log2 normalized mean fold change of dependency probabilities (mutant / control).

• CliffDelta - Cliff's delta non-parametric effect size between mutant and control dependency probabilities. Ranges between -1 to 1.

• dip_pval - Hartigan's dip test p-value. Tests whether distribution of mutant dependency probability is unimodel. If dip test is rejected (p-value < 0.05), this indicates that there is a multimodel dependency probability distribution and that there may be another factor contributing to this separation.

• Interaction_score - Combined value generated from signed p-values: ⁠-log10(Pval) \* sign(log2FC_by_median)⁠

Value

A data frame containing results from the genetic screen. A copy is also saved to the directory defined in output_dir.

Examples

gretta_data_dir <- './GRETTA_example/'
gretta_output_dir <- './GRETTA_example_output/'

if(!dir.exists(gretta_data_dir)){
  download_example_data(".")
}

if(!dir.exists(gretta_data_dir)){
  download_example_data(".")
}

Screen_results <- GI_screen(
control_id = c('ACH-001354', 'ACH-000274', 'ACH-001799'), 
mutant_id = c('ACH-000911', 'ACH-001957', 'ACH-000075'), 
gene_list = c('ARID1A', 'ARID1B', 'SMARCA2'),
core_num = 2, 
output_dir = gretta_output_dir,
data_dir = gretta_data_dir)

ytakemon/GINIR documentation built on Oct. 11, 2024, 6:06 a.m.