R/RemoveSparseFeatures.R

In CountClust: Clustering and Visualizing RNA-Seq Expression Data using Grade of Membership Models

#' Removes features with a lot of 0 counts
#'
#' @description This function deals with zero counts in the counts dataset.
#'               If for a feature, the proportion of zeros across the
#'              samples is greater than filter_prop, then we remove the feature.
#'
#' @param  data count data in a sample by feature matrix.
#' @param  filter_prop threshold proportion. If the proportion of zeros for
#'                      the feature exceeds this threshold then we remove the
#'                      feature altogether. Default is 0.9.
#'
#' @keywords counts data, feature extraction
#'
#' @return  Returns a list with
#'    \item{data}{filtered data with sparse features removed}
#'    \item{sparse_features}{the feature names of the features
#'                           found sparse and removed}
#' @export
#'
#' @examples
#' mat <- rbind(c(2,0,3,0,4),c(4,5,5,0,0),c(30,34,63,25,0),c(0,0,0,0,0));
#' RemoveSparseFeatures(mat, filter_prop = 0.5)
#' RemoveSparseFeatures(mat)
#'
RemoveSparseFeatures <- function(data, filter_prop=0.9)
{
    zero_prop_cols <- apply(data,2,function(x) length(which(x==0))/length(x));

    if(max(zero_prop_cols)<filter_prop){
        message('no features filtered due to sparsity',
                domain = NULL, appendLF = TRUE)
        ll <- list("data" = data,
                   "sparse_features" = as.numeric())
        return(ll)
    } else{
        features_to_remove <- as.numeric(which(zero_prop_cols > filter_prop));
        data_filtered <- as.matrix(data[,-features_to_remove]);
        ll <- list("data" = data_filtered,
                   "sparse_features" = colnames(data)[features_to_remove])
        return(ll)
    }
}