normal_cancer_pairs_from_metadata.12-2-16

normal_cancer_pairs_from_metadata <- function(metadata_file, debug=FALSE){

    ### SUBS
    
    # function to import data and metadata --- better than old import_data as it can hand numerical and nominal data
    import_metadata <- function(group_table){ #, group_column, sample_names){
        metadata_matrix <- as.matrix( # Load the metadata table (same if you use one or all columns)
            read.table(
                file=group_table,row.names=1,header=TRUE,sep="\t",
                colClasses = "character", check.names=FALSE,
                comment.char = "",quote="",fill=TRUE,blank.lines.skip=FALSE
            )
        )
    }
    
    # function to export data
    export_data <- function(data_object, file_name){
        write.table(data_object, file=file_name, sep="\t", col.names = NA, row.names = TRUE, quote = FALSE, eol="\n")
    }

    
    ### MAIN

    # load the metadata file
    metadata_table <- import_metadata(metadata_file)
    if( debug==TRUE ){ TEST.metadata_table <<- metadata_table }

    # find the case ids that are present in more than one sample
    duplicated_case_ids_bool <- duplicated( metadata_table['hits.case_id.1',] )
    duplicated_case_ids <- metadata_table['hits.case_id.1',duplicated_case_ids_bool]
    duplicated_case_ids <- unique(duplicated_case_ids)
    if( debug==TRUE ){ TEST.duplicated_case_ids <<- duplicated_case_ids }

    # make a list of the case ids that have abundance >= 2 # assume that it is a pair (optimize this later)

    output_matrix <- matrix()
    column_names <- vector( mode="character" )
    first_hit <- 0
    #rownames(output_matrix) <- rownames(metadata_table)

    if( debug==TRUE ){ print("Made it here (1)") }
    
    for( i in 1:ncol(metadata_table) ){
        if( debug==TRUE ){ print("Made it here (2.a)") }
        if( debug==TRUE ){ print(i) }
        if( metadata_table['hits.case_id.1',i] %in% duplicated_case_ids ){
            if( debug==TRUE ){ print("Made it here (2.b)") }
            if( debug==TRUE ){ print(paste(metadata_table['hits.case_id.1',i])) }
            if ( first_hit==0){
                output_matrix <- as.matrix( metadata_table[,i], ncol=1, drop=FALSE )
                column_names <- colnames(metadata_table)[i]
                first_hit =+ 1
            }else{
                output_matrix <- cbind( output_matrix, metadata_table[,i] )
                column_names <- c( column_names, colnames(metadata_table)[i])
            }
        }
    }
    colnames( output_matrix ) <- column_names

    if( debug==TRUE ){
        print("Made it here (3)")
        TEST.output_matrix <<- output_matrix
    }
    
    # order columns by case id
    output_matrix <- output_matrix[ , order( output_matrix[ 'hits.case_id.1', ] ) ]

    if( debug==TRUE ){ print("Made it here (4)") }

    # create an additional metdata that has a clearer differenciation between cancer and normal
    simple_descriptors <- vector(mode="character")
    column_count <- 0
    for( i in 1:ncol(output_matrix) ){
        simple_descriptor <- "Cancer"
        if( column_count == 0 ){
            if( grepl("ormal", output_matrix['hits.samples.sample_type.1',i])==TRUE){ simple_descriptor <- "Normal" }
            #if( grepl("ormal", output_matrix['hits.samples.sample_type.2',i])==TRUE){ simple_descriptor <- "Normal" }
            #if( grepl("ormal", output_matrix['hits.samples.sample_type.3',i])==TRUE){ simple_descriptor <- "Normal" }
            simple_descriptors <- simple_descriptor
            column_count =+ 1
        }else{
            if( grepl("ormal", output_matrix['hits.samples.sample_type.1',i])==TRUE){ simple_descriptor <- "Normal" }
            #if( grepl("ormal", output_matrix['hits.samples.sample_type.2',i])==TRUE){ simple_descriptor <- "Normal" }
            #if( grepl("ormal", output_matrix['hits.samples.sample_type.3',i])==TRUE){ simple_descriptor <- "Normal" }
            simple_descriptors <- c(simple_descriptors,simple_descriptor) 
        }
    }
    output_matrix <- rbind(output_matrix, simple_descriptors)
    rownames(output_matrix)[nrow(output_matrix)] <- "C_vs_N"
    
    # export results
    output_filename <- paste(gsub(".txt", "", metadata_file), ".C_N_pairs.txt", sep="", collapse="")
    export_data(output_matrix, output_filename)
    
}







## hits.case_id.1
## hits.samples.sample_type.1
## hits.samples.sample_type.2
## hits.samples.sample_type.3