% Generated by roxygen2: do not edit by hand % Please edit documentation in R/utils.R \name{impute_met_files} \alias{impute_met_files} \title{Impute/predict methylation files} \usage{ impute_met_files( met_dir, outdir = NULL, obj, anno_region, basis = NULL, is_predictive = TRUE, no_cores = NULL ) } \arguments{ \item{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 \code{\link{create_melissa_data_obj}}), where \code{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>.} \item{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.} \item{obj}{Output of Melissa inference object.} \item{anno_region}{Annotation region object. This will be the outpuf of \code{\link{create_melissa_data_obj}} function, e..g melissa_data$anno_region. This is required to select those regions that were used to train Melissa.} \item{basis}{Basis object, if NULL we perform imputation using Melissa, otherwise using BPRMeth (then \code{obj} should be BPRMeth output).} \item{is_predictive}{Logical, use predictive distribution for imputation, or choose the cluster label with the highest responsibility.} \item{no_cores}{Number of cores to be used for parallel processing of data.} } \value{ A new directory \code{outdir} containing files (cells) with predicted / imputed methylation states per CpG location. } \description{ Make predictions of missing methylation states, i.e. perfrom imputation using Melissa. Each file in the directory will be used as input and a new file will be created in \code{outdir} with an additional column containing the predicted met state (value between 0 and 1). Note that predictions will be made only on \code{annotation regions} that were used for training Melissa. Check \code{\link{impute_test_met}}, if you want to make predictions only on test data. } \examples{ \dontrun{ # Met directory met_dir <- "name_of_met_dir" # Annotation file name anno_file <- "name_of_anno_file" # Create data object melissa_data <- create_melissa_data_obj(met_dir, anno_file) # Run Melissa melissa_obj <- melissa(X = melissa_data$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) } } \seealso{ \code{\link{create_melissa_data_obj}}, \code{\link{melissa}}, \code{\link{filter_regions}} } \author{ C.A.Kapourani \email{C.A.Kapourani@ed.ac.uk} }