impute_met_files: Impute/predict methylation files

In andreaskapou/Melissa: Bayesian clustering and imputationa of single cell methylomes

Description

Make predictions of missing methylation states, i.e. perfrom imputation using Melissa. Each file in the directory will be used as input and a new file will be created in outdir with an additional column containing the predicted met state (value between 0 and 1). Note that predictions will be made only on annotation regions that were used for training Melissa. Check impute_test_met, if you want to make predictions only on test data.

Usage

impute_met_files(
  met_dir,
  outdir = NULL,
  obj,
  anno_region,
  basis = NULL,
  is_predictive = TRUE,
  no_cores = NULL
)

Arguments

met_dir	Directory of methylation files, each file corresponds to a single cell. It should contain three columns <chr> <pos> <met_state> (similar to the input required by create_melissa_data_obj), where met_state can be any value that denotes missing CpG information, e.g. -1. Note that files can contain also CpGs for which we have coverage information, and we can check the predictions made by Melissa, hence the value can also be 0 (unmet) or (1) met. Predictions made by Melissa, will not change the <met_state> column. Melissa will just add an additional column named <predicted>.
outdir	Directory to store the output files for each cell with exactly the same name. If NULL, then a directory called 'imputed' inside 'met_dir' will be created by default.
obj	Output of Melissa inference object.
anno_region	Annotation region object. This will be the outpuf of create_melissa_data_obj function, e..g melissa_data$anno_region. This is required to select those regions that were used to train Melissa.
basis	Basis object, if NULL we perform imputation using Melissa, otherwise using BPRMeth (then obj should be BPRMeth output).
is_predictive	Logical, use predictive distribution for imputation, or choose the cluster label with the highest responsibility.
no_cores	Number of cores to be used for parallel processing of data.

Value

A new directory outdir containing files (cells) with predicted / imputed methylation states per CpG location.

Author(s)

C.A.Kapourani [e-mail]

See Also

create_melissa_data_obj, melissa, filter_regions

Examples

## Not run: 
# Met directory
met_dir <- "name_of_met_dir"
# Annotation file name
anno_file <- "name_of_anno_file"
# Create data object
melissa_data <- create_melissa_data_obj(met_dir, anno_file)
# Run Melissa
melissa_obj <- melissa(X = melissa_data$met, K = 2)
# Annotation object
anno_region <- melissa_data$anno_region

# Peform imputation
impute_met_dir <- "name_of_met_dir_for_imputing_cells"
out <- impute_met_files(met_dir = impute_met_dir, obj = melissa_obj,
                        anno_region = anno_region)

## End(Not run)

andreaskapou/Melissa documentation built on June 12, 2020, 5:54 p.m.