Associate factors to covariates of interest — get

Performs Analysis of Variance (ANOVA) or linear models to associate factor scores with covariates of interest provided by the user

Usage

get_associations(
  model,
  metadata,
  sample_id_column,
  test_variable,
  test_type = "categorical",
  categorical_type = "parametric",
  group = FALSE
)

Arguments

model: A MOFA2 model.
metadata: A data frame containing the annotations of the samples included in the MOFA model.
sample_id_column: A string character that refers to the column in metadata where the sample identifier is located.
test_variable: A string character that refers to the column in metadata where the covariate to be tested is located.
test_type: A string character ("categorical", "continuous").
categorical_type: A string character ("parametric", "not_parametric"), only applies for categorical data.
group: Boolean flag TRUE/FALSE, to specify if a grouped MOFA model is provided.

Value

A dataframe in a tidy format containing the p-values of the association tests per factor

Details

Given a covariate of interest and a defined test, this function tests for associations with factor scores. For categorical tests, ANOVAs are fitted, while for continous variables, linear models. P-values are corrected using the Benjamini-Hochberg procedure.

Examples

inputs_dir <- base::system.file("extdata", package = "MOFAcellulaR")
model <- MOFA2::load_model(file.path(inputs_dir, "testmodel.hdf5"))
#> Warning: Factor(s) 1 are strongly correlated with the total number of expressed features for at least one of your omics. Such factors appear when there are differences in the total 'levels' between your samples, *sometimes* because of poor normalisation in the preprocessing steps.
metadata <- readRDS(file.path(inputs_dir, "testmetadata.rds"))
metadata$var <- stats::rnorm(nrow(metadata))

categorical_assoc <- get_associations(model = model,
                                      metadata = metadata,
                                      sample_id_column = "sample",
                                      test_variable = "patient_group",
                                      test_type = "categorical",
                                      group = FALSE)

continuous_assoc <- get_associations(model = model,
                                      metadata = metadata,
                                      sample_id_column = "sample",
                                      test_variable = "var",
                                      test_type = "continuous",
                                      group = FALSE)