view PLIDflow/scripts/clusterfinder_Auto.R @ 6:795e11fac81b draft default tip

Included new tools for standardization
author bitlab
date Wed, 22 Apr 2020 06:12:00 -0400
parents 6fcfa4756040
children
line wrap: on
line source

# #clusterfinder_Auto.R makes a file containig the coordenates of the identified cluster/-s by Silhouette criterion  
# 
#!/usr/bin/env Rscript      
args = commandArgs(trailingOnly=TRUE) 

if(length(args) < 1){
  stop("USE: Rscript clusterfinder_Auto.R <session_id>")
}

#Arguments definition 

session_id <- args[1]

print(session_id)

setwd(session_id)

options(error=traceback)
options(warn=-1)

#Load required library
library(cluster)

#Load required libraries
library(clValid)

#Loading required package: cluster
library(fpc)

#Install NbClust package
#install.packages("NbClust",dependencies = TRUE)

#Loading required package: NbClust
library(NbClust)


#Read the dataset
data_centers <- read.table("fillouts1file.txt", header=TRUE, sep=";", na.strings="") 
#data_centers <- read.table("fillouts1file.txt", header=TRUE, sep=";", na.strings="") 


#Run the XXX for selection of the number of clusters

clustersdata <- NbClust(data_centers, diss=NULL, distance = "euclidean", min.nc=2, max.nc=15, method = "kmeans", index = "all", alphaBeale = 0.1)

#capture.output(clustersdata$Best.nc[1,], file = "bestnumberclusters.txt")

#capture.output(NbClust(data_centers, diss=NULL, distance = "euclidean", min.nc=2, max.nc=15, method = "kmeans", index = "all", alphaBeale = 0.1), file = "proofclusters.txt")

#capture.output(NbClust(data_centers, diss=NULL, distance = "euclidean", min.nc=2, max.nc=15, method = "kmeans", index = "all", alphaBeale = 0.1), file = "numclusters.txt")

#best_num_cluster <- scan("proofclusters.txt", what = character(), quiet = TRUE)

k.max <- as.integer(rownames(table(clustersdata$Best.nc[1,]))[which.max(apply(table(clustersdata$Best.nc[1,]),MARGIN=1,max))])

km.res <- kmeans(as.matrix(data_centers), centers = k.max, nstart = 25)

best_position <- as.integer(rownames(table(clustersdata$Best.nc[1,]))[which.max(apply(table(clustersdata$Best.nc[1,]),MARGIN=1,max))]) 

# Make clustercoordenates table
#X Coordenate
start_x <- best_position + 1
end_x <- start_x + (best_position - 1)

cluster_pos_x <- 1
x_pos <- c()
for (c in start_x:end_x){
	  x_pos[cluster_pos_x] <- km.res$centers[c]
  cluster_pos_x <- cluster_pos_x + 1
}
#print(x_pos)

#Y Coordenate
start_y <- end_x + 1
end_y <- start_y + (best_position - 1)

cluster_pos_y <- 1
y_pos <- c()
for(cc in start_y:end_y){
	  y_pos[cluster_pos_y] <- km.res$centers[cc]
  cluster_pos_y <- cluster_pos_y + 1
}
#print(y_pos)

#Z Coordenate
start_z <- end_y + 1
end_z <- start_z + (best_position - 1)

cluster_pos_z <- 1
z_pos <- c()
for(ccc in start_z:end_z){
	  z_pos[cluster_pos_z] <- km.res$centers[ccc]
  cluster_pos_z <- cluster_pos_z + 1
}
#print(z_pos)

#Create a file with clusters coordenates. Cluster coordenates are vectors x_pos, y_pos, z_pos
num_filas <- length(x_pos)
clusters_tabla <- matrix(1, nrow = num_filas, ncol = 4) #columns are column 1 number of ccluster, column 2 x-coordenates, colum 3 y-coordenates, column 4 z-coordenates

##Add number of cluster located in column 1
for(i in 1:num_filas){
  clusters_tabla[i,1] <- i
}

##Add cluster x,y,z-coordenates in clusters_tabla 
#x-coordenates
v_pos_x <- 1
for(f in 1:num_filas){
  clusters_tabla[f,2] <- x_pos[v_pos_x]
  v_pos_x <- v_pos_x + 1
}

#y-coordenates
v_pos_y <- 1
for(ff in 1:num_filas){
  clusters_tabla[ff,3] <- y_pos[v_pos_y]
  v_pos_y <- v_pos_y + 1
}

#z-coordenates
v_pos_z <- 1
for(fff in 1:num_filas){
  clusters_tabla[fff,4] <- z_pos[v_pos_z]
  v_pos_z <- v_pos_z + 1
}

#Write the head for the file which contains the number of clusters annd their coordenates  
cabecera_clusterscoordenates <- paste("cluster", sep = " ")
cabecera <- c("x","y","z")

for(i in 1:3){
  cabecera_clusterscoordenates <- paste(cabecera_clusterscoordenates, cabecera[i], sep=" ")
}
#write(cabecera_clusterscoordenates, file="/home/galaxy/galaxy/tools/proteindocking/scripts/clusterscoordenates.txt", append= TRUE) 
write(cabecera_clusterscoordenates, file="clusterscoordenates.txt", append= TRUE) 

#Write rows containing number of cluster and coordenates
for(i in 1:nrow(clusters_tabla)){
  fila_completa <- paste(clusters_tabla[i,1], sep = " ")
  for(j in 2:ncol(clusters_tabla)){
    fila_completa <- paste(fila_completa, clusters_tabla[i,j], sep=" ")
  }
  write(fila_completa, file="clusterscoordenates.txt", append= TRUE)
  #write(fila_completa, file="/home/galaxy/galaxy/tools/proteindocking/scripts/clusterscoordenates.txt", append= TRUE)
}


#Table to see Binding Sited finded in Galaxy screen
#BS_screen <- scan("/home/galaxy/galaxy/tools/proteindocking/scripts/clusterscoordenates.txt", what = character(), quiet = TRUE)
BS_screen <- scan("clusterscoordenates.txt", what = character(), quiet = TRUE)

BS_table <- matrix(1, nrow = nrow(clusters_tabla), ncol = ncol(clusters_tabla))

bs_count <- 5
for(i in 1:nrow(clusters_tabla)){
  for(j in 1:ncol(clusters_tabla)){
    BS_table[i,j] <- BS_screen[bs_count]
    bs_count <- bs_count +1
  }
}

colnames(BS_table) <- c("Binding Site", "X", "Y", "Z")
print(BS_table)