annotate cluster_ceamarc.R @ 0:92ca83b60089 draft

planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
author ecology
date Wed, 11 Sep 2024 09:18:48 +0000
parents
children c9109f90d229
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
1 ##13/04/2023
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
2 ##Seguineau Pauline
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
3 ### Clustering with Clara algorithm
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
4
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
5 #load library
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
6 library(cluster)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
7 library(dplyr)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
8 library(tidyverse)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
9
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
10 #load arguments
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
11 args = commandArgs(trailingOnly=TRUE)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
12 if (length(args)==0)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
13 {
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
14 stop("This tool needs at least one argument")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
15 }else{
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
16 data <- args[1]
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
17 enviro <- args[2]
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
18 data.bio <- args[3]
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
19 k <- as.numeric(args[4])
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
20 metric <- args[5]
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
21 sample <- as.numeric(args[6])
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
22 }
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
23
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
24 #load data
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
25 env.data <- read.table(enviro, header=TRUE, sep="\t",dec = ".", na.strings = "-9999")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
26 data.bio <- read.table(data.bio, header=TRUE, sep="\t")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
27 test3 <- read.table(data, header = TRUE, sep="\t")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
28
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
29 ######################################################################################################
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
30 #Make clustering
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
31
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
32 k <- k #number of clusters
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
33 test5 <- clara(test3, k, metric = metric, samples = sample, sampsize = min(nrow(test3), (nrow(data.bio)/nrow(test3))+2*k))
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
34
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
35 #######################################################################################################
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
36 #save results
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
37
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
38 png("sih.png")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
39 plot(silhouette(test5))
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
40 dev.off()
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
41
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
42 clus <- cbind(data.bio[1:nrow(test3), 1:2],test5$clustering)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
43 names(clus) <- c("lat", "long", "cluster")
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
44 clus <- cbind(clus,test3,env.data[,3:19])
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
45
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
46 write.table(clus[1:3], file = "points_clus.tsv", quote = FALSE, sep="\t", row.names = FALSE)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
47 write.table(clus, file = "clus.tsv", quote = FALSE, sep="\t", row.names = FALSE)
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
48
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
49
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
50
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
51
92ca83b60089 planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/Ecoregionalization_workflow commit e03df85746a3b61a382a5ee7e3357a8bf42a5097
ecology
parents:
diff changeset
52