mmahmoudian · September 9, 2022 16:41
diff --git a/venn_zones.R b/venn_zones.R
 venn_zones <- function(lst, output = "data.frame", verbose = TRUE){
    ## Description:
    ##  A function to get the items in each zone of a venn diagram when
    ##   provided with the exact list that was given to a venn::venn().
    ##
    ## Arguments:
    ##  lst:     The exact list that was given to a venn::venn()
    ##  output:  The format that the output should be returned. Valid options are
    ##            "data.frame" and "list".
    ##  verbose: If the function should show which zone it is working on
    ##
    ## Example:
    ##  set.seed(12345)
    ##  a <- list(list1 = sample(x = letters, size = 20),
    ##            list2 = sample(x = letters, size = 20),
    ##            list3 = sample(x = letters, size = 20))
    ##  res <- venn_zones(a)
    ##  comment(res)



    # create all the combinations
    combinations <- lapply(seq_along(lst), function(x){ combn(x = seq_along(lst), m = x, simplify = F) })


    # create empty objects to be filled in the following for loop
    collective_list <- list()
    collective_names <- c()


    # iterate through the major combinations
    for(i in seq_along(combinations)){
        tmp1 <- combinations[[i]]

        # iterate through the minor combinations
        for(j in seq_along(tmp1)){
            tmp2 <- tmp1[[j]]

            if(is.null(names(lst))){
                tmp_name <- tmp2
            }else{
                tmp_name <- names(lst)[tmp2]
            }

            if(verbose){
                print(tmp_name)
            }

            # create the name of the zone and append it to the collective variable
            collective_names <- c(collective_names, paste0("only.in.", paste(tmp_name, collapse = "&")))

            # get the intersect of the selected sets (this might include some that are also belonging to other sets)
            if(length(tmp_name) == 1){
                tmp_pool <- unlist(lst[tmp_name])
            }else{
                tmp_pool <- intersect2(lst[tmp_name])
            }

            # subtract the items that are not specific to the sets we are intersecting
            tmp_res <- setdiff(tmp_pool, unique(unlist( lst[names(lst)[-tmp2]] )))

            # append to collective variable
            collective_list <- lappend(collective_list, tmp_res)
        }
    }


    # apply the names to the list
    names(collective_list) <- collective_names

    #-------[ prepare output ]-------#
    {
        if(output == "data.frame"){
            # convert the list to dataframe with NA for filling
            result <- list2df(l = collective_list, byrow = F, stringsAsFactors = F, filler = NA)

        }else if(output == "list"){
            result <- collective_list
        }

        # clarify the names through the comments
        comment(result) <- paste("This file contains the items in each zone of the venn diagram (one column per zone).",
                                 "As for the column names we have used the numerical value in front of the following",
                                 "to make the column names easeir to read:",
                                 paste("[", seq_along(names(lst)), "]", names(lst), collapse = "\n"),
                                 sep = "\n")
    }


    return(result)
 }
	venn_zones <- function(lst, output = "data.frame", verbose = TRUE){
	## Description:
	## A function to get the items in each zone of a venn diagram when
	## provided with the exact list that was given to a venn::venn().
	##
	## Arguments:
	## lst: The exact list that was given to a venn::venn()
	## output: The format that the output should be returned. Valid options are
	## "data.frame" and "list".
	## verbose: If the function should show which zone it is working on
	##
	## Example:
	## set.seed(12345)
	## a <- list(list1 = sample(x = letters, size = 20),
	## list2 = sample(x = letters, size = 20),
	## list3 = sample(x = letters, size = 20))
	## res <- venn_zones(a)
	## comment(res)



	# create all the combinations
	combinations <- lapply(seq_along(lst), function(x){ combn(x = seq_along(lst), m = x, simplify = F) })


	# create empty objects to be filled in the following for loop
	collective_list <- list()
	collective_names <- c()


	# iterate through the major combinations
	for(i in seq_along(combinations)){
	tmp1 <- combinations[[i]]

	# iterate through the minor combinations
	for(j in seq_along(tmp1)){
	tmp2 <- tmp1[[j]]

	if(is.null(names(lst))){
	tmp_name <- tmp2
	}else{
	tmp_name <- names(lst)[tmp2]
	}

	if(verbose){
	print(tmp_name)
	}

	# create the name of the zone and append it to the collective variable
	collective_names <- c(collective_names, paste0("only.in.", paste(tmp_name, collapse = "&")))

	# get the intersect of the selected sets (this might include some that are also belonging to other sets)
	if(length(tmp_name) == 1){
	tmp_pool <- unlist(lst[tmp_name])
	}else{
	tmp_pool <- intersect2(lst[tmp_name])
	}

	# subtract the items that are not specific to the sets we are intersecting
	tmp_res <- setdiff(tmp_pool, unique(unlist( lst[names(lst)[-tmp2]] )))

	# append to collective variable
	collective_list <- lappend(collective_list, tmp_res)
	}
	}


	# apply the names to the list
	names(collective_list) <- collective_names

	#-------[ prepare output ]-------#
	{
	if(output == "data.frame"){
	# convert the list to dataframe with NA for filling
	result <- list2df(l = collective_list, byrow = F, stringsAsFactors = F, filler = NA)

	}else if(output == "list"){
	result <- collective_list
	}

	# clarify the names through the comments
	comment(result) <- paste("This file contains the items in each zone of the venn diagram (one column per zone).",
	"As for the column names we have used the numerical value in front of the following",
	"to make the column names easeir to read:",
	paste("[", seq_along(names(lst)), "]", names(lst), collapse = "\n"),
	sep = "\n")
	}


	return(result)
	}