dce-methods: Differential Causal Effects - main function

In kpj/dce: Pathway Enrichment Based on Differential Causal Effects

Differential Causal Effects - main function

Description

Main function to compute differential causal effects and the pathway enrichment

Usage

dce(
  graph,
  df_expr_wt,
  df_expr_mt,
  solver = "lm",
  solver_args = list(),
  adjustment_type = "parents",
  effect_type = "total",
  p_method = "hmp",
  test = "wald",
  lib_size = FALSE,
  deconfounding = FALSE,
  conservative = FALSE,
  log_level = logger::INFO
)

## S4 method for signature 'igraph'
dce(
  graph,
  df_expr_wt,
  df_expr_mt,
  solver = "lm",
  solver_args = list(),
  adjustment_type = "parents",
  effect_type = "total",
  p_method = "hmp",
  test = "wald",
  lib_size = FALSE,
  deconfounding = FALSE,
  conservative = FALSE,
  log_level = logger::INFO
)

## S4 method for signature 'graphNEL'
dce(
  graph,
  df_expr_wt,
  df_expr_mt,
  solver = "lm",
  solver_args = list(),
  adjustment_type = "parents",
  effect_type = "total",
  p_method = "hmp",
  test = "wald",
  lib_size = FALSE,
  deconfounding = FALSE,
  conservative = FALSE,
  log_level = logger::INFO
)

## S4 method for signature 'matrix'
dce(
  graph,
  df_expr_wt,
  df_expr_mt,
  solver = "lm",
  solver_args = list(),
  adjustment_type = "parents",
  effect_type = "total",
  p_method = "hmp",
  test = "wald",
  lib_size = FALSE,
  deconfounding = FALSE,
  conservative = FALSE,
  log_level = logger::INFO
)

Arguments

graph	valid object defining a directed acyclic graph
df_expr_wt	data frame with wild type expression values
df_expr_mt	data from with mutation type expression values
solver	character with name of solver function
solver_args	additional arguments for the solver function. please adress this argument, if you use your own solver function. the default argument works with glm functions in the packages MASS, stats and glm2
adjustment_type	character string for the method to define the adjustment set Z for the regression
effect_type	method of computing causal effects
p_method	character string. "mean", "sum" for standard summary functions, "hmp" for harmonic mean or any method from package 'metap', e.g., "meanp" or "sump".
test	either "wald" for testing significance with the wald test or "lr" for using a likelihood ratio test. Alternatively, "vcovHC" can improve results for zero-inflated date, i.e., from single cell RNAseq experiments.
lib_size	either a numeric vector of the same length as the sum of wild type and mutant samples or a logical. If TRUE, it is recommended that both data sets include not only the genes included in the graph but all genes available in the original data set.
deconfounding	indicates whether adjustment against latent confounding is used. If FALSE, no adjustment is used, if TRUE it adjusts for confounding by automatically estimating the number of latent confounders. The estimated number of latent confounders can be chosen manually by setting this variable to some number.
conservative	logical; if TRUE, does not use the indicator variable for the variables in the adjustment set
log_level	Control verbosity (logger::INFO, logger::DEBUG, ...)

Value

list of matrices with dces and corresponding p-value

Examples

dag <- create_random_DAG(30, 0.2)
X.wt <- simulate_data(dag)
dag.mt <- resample_edge_weights(dag)
X.mt <- simulate_data(dag)
dce(dag,X.wt,X.mt)

kpj/dce documentation built on Oct. 29, 2022, 1:40 a.m.