2021-08-25 15:01:18 +02:00
|
|
|
library(data.table)
|
|
|
|
|
|
2021-09-30 12:54:40 +02:00
|
|
|
#' Perform a cluster analysis.
|
|
|
|
|
#'
|
|
|
|
|
#' This function will cluster the data using `hclust` and `cutree` (with the
|
|
|
|
|
#' specified height). Every cluster with at least two members qualifies for
|
|
|
|
|
#' further analysis. Clusters are then ranked based on their size in relation
|
2021-10-04 09:19:38 +02:00
|
|
|
#' to the total number of possible values (`n`). The return value is a final
|
|
|
|
|
#' score between zero and one. Lower ranking clusters contribute less to this
|
|
|
|
|
#' score.
|
|
|
|
|
clusteriness <- function(data, n, height = 1000000) {
|
2021-10-09 14:22:28 +02:00
|
|
|
# Return a score of 0.0 if there is just one or no value at all.
|
|
|
|
|
if (length(data) < 2) {
|
|
|
|
|
return(0.0)
|
|
|
|
|
}
|
|
|
|
|
|
2021-09-30 12:54:40 +02:00
|
|
|
# Cluster the data and compute the cluster sizes.
|
|
|
|
|
|
|
|
|
|
tree <- hclust(dist(data))
|
|
|
|
|
clusters <- cutree(tree, h = height)
|
|
|
|
|
cluster_sizes <- sort(tabulate(clusters), decreasing = TRUE)
|
|
|
|
|
|
2021-10-09 14:22:28 +02:00
|
|
|
# Compute the "clusteriness" score.
|
2021-09-30 12:54:40 +02:00
|
|
|
|
|
|
|
|
score <- 0.0
|
|
|
|
|
|
|
|
|
|
for (i in seq_along(cluster_sizes)) {
|
|
|
|
|
cluster_size <- cluster_sizes[i]
|
|
|
|
|
|
|
|
|
|
if (cluster_size >= 2) {
|
|
|
|
|
cluster_score <- cluster_size / n
|
|
|
|
|
score <- score + cluster_score / i
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
score
|
|
|
|
|
}
|
|
|
|
|
|
2021-09-18 23:10:52 +02:00
|
|
|
#' Process genes clustering their distance to telomeres.
|
2021-08-25 15:01:18 +02:00
|
|
|
#'
|
2021-09-18 23:10:52 +02:00
|
|
|
#' The return value will be a data.table with the following columns:
|
|
|
|
|
#'
|
|
|
|
|
#' - `gene` Gene ID of the processed gene.
|
2021-10-09 14:22:28 +02:00
|
|
|
#' - `clusteriness` Score quantidying the gene's clusters.
|
2021-08-25 15:01:18 +02:00
|
|
|
#'
|
2021-09-16 00:06:54 +02:00
|
|
|
#' @param distances Gene distance data to use.
|
2021-08-29 13:25:12 +02:00
|
|
|
#' @param species_ids IDs of species to include in the analysis.
|
2021-09-16 00:06:54 +02:00
|
|
|
#' @param gene_ids Genes to include in the computation.
|
2021-09-18 23:10:52 +02:00
|
|
|
process_clustering <- function(distances, species_ids, gene_ids) {
|
2021-09-16 00:06:54 +02:00
|
|
|
results <- data.table(gene = gene_ids)
|
2021-10-09 14:22:28 +02:00
|
|
|
species_count <- length(species)
|
2021-09-21 16:47:13 +02:00
|
|
|
|
2021-10-09 14:22:28 +02:00
|
|
|
# Prefilter the input data by species.
|
|
|
|
|
distances <- distances[species %chin% species_ids]
|
2021-08-25 15:01:18 +02:00
|
|
|
|
2021-10-09 14:22:28 +02:00
|
|
|
# Add an index for quickly accessing data per gene.
|
|
|
|
|
setkey(distances, gene)
|
2021-08-25 15:01:18 +02:00
|
|
|
|
2021-10-09 14:22:28 +02:00
|
|
|
#' Perform the cluster analysis for one gene.
|
|
|
|
|
compute <- function(gene_id) {
|
|
|
|
|
clusteriness(distances[gene_id, distance], species_count)
|
2021-08-25 15:01:18 +02:00
|
|
|
}
|
|
|
|
|
|
2021-10-09 14:22:28 +02:00
|
|
|
results[, clusteriness := compute(gene), by = 1:nrow(results)]
|
2021-08-25 15:01:18 +02:00
|
|
|
}
|