sys_home<-Sys.getenv("HOME")
if(sys_home=="/Users/prosikie"){home_path<-"/Users/prosikie/Desktop/new_december"}
if(sys_home=="/home/mrosikie"){home_path<-"/home/mrosikie"}

project_path<-file.path(home_path,"documentation","meta_suez")
setwd(project_path)

#file with result of rdp classification with EzbioCloud database
rdp_classif_file<-file.path(project_path,"vsearch_data_November2017","rdp","meta_suez_EzBioCloud_centr_vsearch_rep_seq_2_97","meta_suez_centr_vsearch_rep_seq_2_97_tax_assignments.txt")
rdp_table<-read.table(rdp_classif_file,as.is=T,sep="\t")

#creating tax table
get_tax<-function(x) {
  tax_split<-strsplit(x,";")[[1]]
  N_tax_fields<-length(tax_split)
  tax_vector<-c(tax_split,rep(NA,7-N_tax_fields))
  return(tax_vector)
}

tax_table<-matrix(ncol=7,nrow=nrow(rdp_table))
for (xx in 1:nrow(rdp_table)){
  tax_table[xx,]<-get_tax(rdp_table[xx,2])
}

colnames(tax_table)<-c("kingdom","phylum","class","order","family","genus","species")
rownames(tax_table)<-rdp_table[,1]

#file with otus abundance
output_folder<-file.path(project_path,"vsearch_data_November2017","OTU_tables_vsearch_centr_min2_97")
load(file.path(output_folder,"meta_suez_OTU_search_table_centr_2_97.RData"))
otus_table<-OTU_search_table_2_97
rm(OTU_search_table_2_97)

#############################################
general_table_folder<-file.path(output_folder,"otus_tax_tables_vsearch_centr_min2_97")
if(!file.exists(general_table_folder)){dir.create(general_table_folder)}

general_table_folder_plus<-file.path(output_folder,"otus_tax_tables_plus_vsearch_centr_min2_97")
if(!file.exists(general_table_folder_plus)){dir.create(general_table_folder_plus)}

general_table_folder_pr<-file.path(output_folder,"otus_tax_tables_pr_vsearch_centr_min2_97")
if(!file.exists(general_table_folder_pr)){dir.create(general_table_folder_pr)}

Bacteria_table_folder<-file.path(output_folder,"otus_Bacteria_tax_tables_vsearch_centr_min2_97")
if(!file.exists(Bacteria_table_folder)){dir.create(Bacteria_table_folder)}

Bacteria_table_folder_plus<-file.path(output_folder,"otus_Bacteria_tax_tables_plus_vsearch_centr_min2_97")
if(!file.exists(Bacteria_table_folder_plus)){dir.create(Bacteria_table_folder_plus)}

Bacteria_table_folder_pr<-file.path(output_folder,"otus_Bacteria_tax_tables_pr_vsearch_centr_min2_97")
if(!file.exists(Bacteria_table_folder_pr)){dir.create(Bacteria_table_folder_pr)}

#####################################
#sorting
tax_table<-tax_table[rownames(otus_table),]
tax_table2<-cbind(OTU_name=rownames(tax_table),tax_table)
write.csv(tax_table2,file=file.path(project_path,"vsearch_data_November2017","meta_suez_vsearch_centr_min2_97_tax.csv"),quote=F,row.names=F)

#getting tables with all kingdoms 
kingdom_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'kingdom']),sum)})
species_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'species']),sum)})
genus_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'genus']),sum)})
family_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'family']),sum)})
class_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'class']),sum)})
order_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'order']),sum)})
phylum_table<-apply(otus_table,2,function(x){sapply(split(x,tax_table[,'phylum']),sum)})

for (xx in c("otus","species","genus","family","class","order","phylum","kingdom")){
  write.csv(get(paste(xx,"_table",sep="")),file=file.path(general_table_folder,paste(xx,"_table.csv",sep="")),quote=F)
}

for (xx in c("otus","species","genus","family","class","order","phylum")){
  assign(paste(xx,"_table_plus",sep=""),rbind(get(paste(xx,"_table",sep="")),Unclassified=colSums(otus_table)-colSums(get(paste(xx,"_table",sep="")))))
  write.csv(get(paste(xx,"_table_plus",sep="")),file=file.path(general_table_folder_plus,paste(xx,"_table_plus.csv",sep="")),quote=F)
}

for (xx in c("otus","species","genus","family","class","order","phylum")){
  plus_table<-get(paste(xx,"_table_plus",sep=""))
  assign(paste(xx,"_table_plus_pr",sep=""),t(t(plus_table)/colSums(plus_table))*100)
  write.csv(get(paste(xx,"_table_plus_pr",sep="")),file=file.path(general_table_folder_pr,paste(xx,"_table_plus_pr.csv",sep="")),quote=F)
}

kingdom_table_pr<-t(t(kingdom_table)/colSums(otus_table))*100
write.csv(kingdom_table_pr,file=file.path(general_table_folder_pr,paste("kingdom_table_pr.csv",sep="")),quote=F)

#Bacteria only 
B_otus_table<-otus_table[tax_table[,1]=="Bacteria",]
B_tax_table<-tax_table[tax_table[,1]=="Bacteria",]

B_species_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'species']),sum)})
B_genus_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'genus']),sum)})
B_family_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'family']),sum)})
B_class_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'class']),sum)})
B_order_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'order']),sum)})
B_phylum_table<-apply(B_otus_table,2,function(x){sapply(split(x,B_tax_table[,'phylum']),sum)})


for (xx in c("B_otus","B_species","B_genus","B_family","B_class","B_order","B_phylum")){
  write.csv(get(paste(xx,"_table",sep="")),file=file.path(Bacteria_table_folder,paste(xx,"_table.csv",sep="")),quote=F)
}

for (xx in c("B_otus","B_species","B_genus","B_family","B_class","B_order","B_phylum")){
  assign(paste(xx,"_table_plus",sep=""),rbind(get(paste(xx,"_table",sep="")),Unclassified=colSums(B_otus_table)-colSums(get(paste(xx,"_table",sep="")))))
  write.csv(get(paste(xx,"_table_plus",sep="")),file=file.path(Bacteria_table_folder_plus,paste(xx,"_table_plus.csv",sep="")),quote=F)
}

for (xx in c("B_otus","B_species","B_genus","B_family","B_class","B_order","B_phylum")){
    plus_table<-get(paste(xx,"_table_plus",sep=""))
    assign(paste(xx,"_table_plus_pr",sep=""),t(t(plus_table)/colSums(plus_table))*100)
    write.csv(get(paste(xx,"_table_plus_pr",sep="")),file=file.path(Bacteria_table_folder_pr,paste(xx,"_table_plus_pr.csv",sep="")),quote=F)
}