R/targetScan.R

In fischuu/hoardeR: Collect and Retrieve Annotation Data for Various Genomic Data Using Different Webservices

extractTSinfo <- function(x){
  ortholog <- strsplit(x,"target=new>")
  ortholog <- strsplit(ortholog[[1]][2],"</a></td>")[[1]][1]
  
  geneName <- strsplit(x,"<td nowrap>")
  geneName <- strsplit(geneName[[1]][2],"</td>")[[1]][1]
    
  totalSites <- strsplit(x,"<B>") 
  consSites <- strsplit(totalSites[[1]][2],"</B>")[[1]][1]
  poorlySites <- strsplit(totalSites[[1]][3],"</B>")[[1]][1]   
  
  res <- data.frame(Ortholog=ortholog,
                    geneName=geneName,
                    consSites=consSites,
                    poorlySites=poorlySites)
  res
}

targetScan <- function(mirna=NULL, species=NULL, release="7.1", maxOut=NULL){
  if(length(mirna)>1){
   out <- mapply(targetScan.internal, mirna=mirna, species=species, release=release, maxOut=maxOut, SIMPLIFY=FALSE) 
  } else{
    out <- targetScan.internal(mirna=mirna, species=species, release=release, maxOut=maxOut)    
  }
  out
}

targetScan.internal <- function(mirna, species, release, maxOut){

# Starting values for later tests
  origMirna <- mirna
  notFound <- FALSE
  allChecked <- FALSE
  
 if(is.null(species)){
  # Guessing species from mirna
    if(substr(mirna,1,3)=="hsa") species <- "Human"
    if(substr(mirna,1,3)=="mmu") species <- "Mouse"
    if(substr(mirna,1,3)=="rno") species <- "Rat"
    if(substr(mirna,1,3)=="ptr") species <- "Chimpanzee"
    if(substr(mirna,1,3)=="mml") species <- "Rhesus"
    if(substr(mirna,1,3)=="bta") species <- "Cow"
    if(substr(mirna,1,3)=="cfa") species <- "Dog"
    if(substr(mirna,1,3)=="mdo") species <- "Opossum"
    if(substr(mirna,1,3)=="gga") species <- "Chicken"
    if(substr(mirna,1,3)=="xtr") species <- "Frog"
  }
  
# Input checks
  if(is.null(mirna)) stop("No mirne name given. Use e.g. 'miR-9-5p'.")
  species <- paste(toupper(substr(species, 1, 1)), tolower(substr(species, 2, nchar(species))), sep="")
  species <- match.arg(species, c("Human", "Mouse", "Rat", "Chimpanzee", "Rhesus", "Cow", "Dog", "Opossum", "Chicken", "Frog"))
  release <- gsub("\\.","",release)
 
# Retrieve the content 
  tsAddress <- paste("http://www.targetscan.org/cgi-bin/targetscan/vert_71/targetscan.cgi?species=",species,"&mirg=",mirna,sep="")
  tsOut <- scan(tsAddress, what = "", sep = "\n", quiet = TRUE)

# Check first if the miRNA is in the targetScan database  
  if(sum(grepl("is not in our miRNA database",tsOut[1:min(100,length(tsOut))]))>0){
       warning(mirna," is not in the targetScan database!\n")
       res <- data.frame(Ortholog=NULL,
                         geneName=NULL,
                         consSites=NULL,
                         poorlySites=NULL)
  }else {
      # Check first, if the targetScan result is unique
        if(sum(grepl("matches multiple families in our miRNA database",tsOut[1:min(100,length(tsOut))]))>0){
          notFound <- TRUE
          multFams <- tsOut[grepl("=miR",tsOut)]
          newMirnas <- character(length(multFams))
          for(i in 1:length(multFams)){
            temp <- strsplit(multFams[i], "</A")[[1]][1]
            newMirnas[i] <- strsplit(temp,'\">')[[1]][2]
          }
          warning("Multiple matches multiple families in the targetScan database for ",mirna,":\n",paste(newMirnas,collapse="; "),"\nOnly the first one is used!")
        # Take the first unique set of miRNAs
          temp <- which(grepl("/", newMirnas)==TRUE)
          takeThis <- 1
          if(length(temp)>0) takeThis <- temp + 1 
      
        # It can happen that none of the derivatives of a miRNA is in the target database, fetch this case here
          while(notFound & !allChecked){
          # Start to check the first reasonable miRNA
            mirna <- newMirnas[takeThis]  
            tsAddress <- paste("http://www.targetscan.org/cgi-bin/targetscan/vert_71/targetscan.cgi?species=",species,"&mirg=",mirna,sep="")
            tsOut <- scan(tsAddress, what = "", sep = "\n", quiet = TRUE)
          # If this is in the database, stop the searching and mark as found, if not go on until all possibilities are checked
            if(sum(grepl("is not in our miRNA database",tsOut[1:min(100,length(tsOut))]))>0){
              if(takeThis < length(newMirnas)){
                takeThis <- takeThis + 1
              } else {
                allChecked <- TRUE
              }
            } else {
              notFound <- FALSE
            }
          }
        }
      # If none of the miRNA derivatis was found in the database, return an emtpy result.
       if(notFound){
         warning(mirna," is not in the targetScan database!\nAlso none of its derivatives",paste(newMirnas,collapse=","),"could be found")
         res <- data.frame(Ortholog=NULL,
                           geneName=NULL,
                           consSites=NULL,
                           poorlySites=NULL)
       } else {
         # Find the rows of interest (Assume it to be in the first 100 rows, if this isn't the case extent the search area)
         startRow <- grepl("<th>total</th>",tsOut[1:min(100,length(tsOut))])
         if(sum(startRow)!=1)  startRow <- grepl("<th>total</th>",tsOut)
         if(sum(startRow)!=1) stop("ERROR: No table provided by targetScan.org!")
         startRow <- which(startRow==1)
         ifelse(is.null(maxOut), maxOut <- length(tsOut)-1, maxOut <- startRow + maxOut - 1)
         
         
         # Now extract the information and put them into a dataframe
         
         # The first row is a bit different, as it is contained in the header row, all others are then standardized
         temp1 <- strsplit(tsOut[startRow],"<td>")
         firstEntry <- paste(temp1[[1]][2],"<td>",temp1[[1]][3],sep="")
         res <- extractTSinfo(firstEntry)
         for(i in (startRow+1):maxOut){
           res <- rbind(res,extractTSinfo(tsOut[i]))
         }         
       }
  }
  res
}