R/Match_query_ccl_with_database.R

In Uniquorn: Identification of cancer cell lines based on their weighted mutational/ variational fingerprint

#' match_query_ccl_to_database
#' 
#' Matches query ccl to the database
#' 
#' @param g_query IRanges object that contains the variants
#' @param ref_gen Reference genome version. All training sets are 
#'  associated with a reference genome version. Default: GRCH37
#' @param library_name a character string giving the name of the library
#' @param mutational_weight_inclusion_threshold a numerical giving
#'  the lower bound for mutational weight to be included
#' @importFrom IRanges subsetByOverlaps
#' @return The R Table sim_list which contains the CoSMIC CLP fingerprints 
match_query_ccl_to_database = function(
    g_query,
    ref_gen = "GRCH37",
    library_name,
    mutational_weight_inclusion_threshold
){
    
    g_mat = read_mutation_grange_objects(
        ref_gen = ref_gen,
        library_name = library_name,
        mutational_weight_inclusion_threshold = 
            mutational_weight_inclusion_threshold,
        test_mode = FALSE
    )
    
    fo_query = GenomicRanges::findOverlaps(
        query = g_query,
        subject = g_mat,
        select = "arbitrary",
        type = "equal"
    )
    match = IRanges::subsetByOverlaps(g_mat, g_query)
    cl_ids = as.character(unlist(
        str_split(GenomicRanges::mcols(match)$Member_CCLs, pattern = "," )
    ))
    cl_ids = str_replace_all(cl_ids,
                            pattern = paste("_",library_name,sep = ""),"" )
    
    matching_variants = sort(table(cl_ids), decreasing = TRUE)
    cl_ids_all = str_replace_all( 
        GenomicRanges::mcols(g_mat)$Member_CCLs, 
        pattern = paste("_",library_name,sep = ""),
        ""
    )
    cl_ids_all = unique(as.character(
        unlist(str_split(cl_ids_all,pattern = ","))))
    missing_ccls = table(
        cl_ids_all[ !( cl_ids_all %in% names(matching_variants))]
    )
    missing_ccls = missing_ccls - 1
    matching_variants = c(matching_variants, missing_ccls)
    
    matching_variants_t = data.frame(
        CCL = names(matching_variants),
        Matches = as.character(matching_variants),
        Library = rep(library_name, length(matching_variants))
    )
    
    ### add stats info
    
    cl_data <<-  show_contained_ccls( verbose = FALSE)
    cl_data = cl_data[cl_data$Library == library_name,]
    
    switch(as.character(mutational_weight_inclusion_threshold),
        "0" = { weight_name = "W0"  },
        "0.25" = { weight_name = "W25"  },
        "0.5" = { weight_name = "W05"  },
        "1" = { weight_name = "W1"  },
        stop(paste( "Could not recocgnize mutational weight ", 
            mutational_weight_inclusion_threshold))
    )
    
    all_muts = as.character(cl_data[[weight_name]])
    ccl_match = match( as.character(
        matching_variants_t$CCL), as.character(cl_data$CCL), nomatch = 0)
    
    missing_libraries = unique(
        as.character(matching_variants_t$Library[ccl_match == 0]))
    
    if(length(missing_libraries) > 0){
        
        for (mis_lib_name in  missing_libraries){
            
            message("Updating database for library: ", mis_lib_name)
            mis_mat = read_mutation_grange_objects(
                ref_gen = ref_gen,
                library_name = mis_lib_name,
                mutational_weight_inclusion_threshold = 
                    mutational_weight_inclusion_threshold
            )
            write_w0_and_split_w0_into_lower_weights(
                g_mat = mis_mat,
                library_name = mis_lib_name,
                ref_gen = ref_gen
            )
        }
        
        cl_data <<-  show_contained_ccls( ref_gen = ref_gen, verbose = FALSE)
        cl_data = cl_data[cl_data$Library == library_name,]
        
        switch(as.character(mutational_weight_inclusion_threshold),
            "0" = { weight_name = "W0"  },
            "0.25" = { weight_name = "W25"  },
            "0.5" = { weight_name = "W05"  },
            "1" = { weight_name = "W1"  },
            stop("Could not recocgnize mutational weight ", 
                mutational_weight_inclusion_threshold)
        )
        
        all_muts <<- as.character(cl_data[[weight_name]])
        ccl_match <<- match( as.character(
            matching_variants_t$CCL
        ), as.character(cl_data$CCL), nomatch = 0)
    }
    
    matching_variants_t$All_variants = all_muts[ccl_match]
    
    return(matching_variants_t)
}