R/addMissingSamples.R

In ss-lab-cancerunit/SpiderSeqR: A Tool for Integration of Big Bio Data

#' Add missing samples from GSEs and SRPs within a data frame
#' 
#' @param df Data frame (must be in searchForAccession 
#' or searchAnywhere output format)
#' @param values A vector (length 2) indicating what will be the values 
#' assigned to the original df rows and to added rows respectively
#' @return Data frame with added missing rows from the same GSE/SRP. 
#' It will also contain an additional column to indicate whether the row 
#' was originally present or not
#' @family Workflow functions
#' 
#' @description 
#' This function is intended to facilitate workflow with the results of 
#' \code{searchAnywhere()} function. The main motivation is to provide missing 
#' samples that did not match the query criteria, but might be useful 
#' when searching for controls 
#' 
#' @examples 
#' startSpiderSeqRDemo()
#' addMissingSamples(searchForAccession("SRR6823653"))
#' # NOTE: searchForAccession is used here to provide the data frame; 
#' # in reality the data frame can be generated by most other 
#' # SpiderSeqR functions (including search functions)
#' 
#' 
#' 
#' @export
#' 
addMissingSamples <- function(df, values = c(1,0)){
    
    # General logic:
    # Check format - whether it adheres to all the column names etc
    # Add column to original df - OTH_Ori
    # Search for accession SRP
    # Bind
    # Split, unlist and unique GSEs
    # Search for accession GSEs
    # Bind 
    
    
    .checkValidColumns(df) # Check that column names are within allowed set
    
    if ( (!"study_accession" %in% colnames(df)) | 
            (!"series_id" %in% colnames(df))){
        
        stop("Missing required columns: study_accession and/or series_id")
        
    }
    
    df$OTH_sample <- values[1] # Decide on the column name ===*===
    
    print(dim(df))
    
    srps <- unique(df$study_accession)
    srps <- srps[!is.na(srps)] # Remove NAs
    
    if (length(srps)>0){
        srp_df <- searchForAccession(srps, file_output = FALSE)
        srp_df <- .unifyDFFormat(srp_df)
        srp_df$OTH_sample <- values[2]
        
        # Remove rows that are already present
        srp_df <- dplyr::anti_join(srp_df, df, by = colnames(df)) 
        
        # Select only corresponding columns
        srp_df <- srp_df[, colnames(df)] 
        
        df <- rbind(df, srp_df)
    }
    
    
    gses <- unique(unlist(strsplit(df$series_id, split = ",")))
    gses <- gses[!is.na(gses)] # Remove NAs
    
    if (length(gses)>0){
        gse_df <- searchForAccession(gses)
        gse_df <- .unifyDFFormat(gse_df)
        gse_df$OTH_sample <- values[2]
        
        # Remove rows that are already present
        gse_df <- dplyr::anti_join(gse_df, df, by = colnames(df)) 
        
        # Select only corresponding columns
        gse_df <- gse_df[, colnames(df)] 
        
        df <- rbind(df, gse_df)
    }
    
    
    col_number <- dim(df)[2]
    
    # Put the added column as the first
    df <- df[ ,c(col_number, seq(1, col_number-1))] 
    
    return(df)
    
}