#!/usr/bin/groovy
// Given a list of pubmed IDS, use NCBO annotator to generate Disease Ontology annotations.
// See http://www.bioontology.org/wiki/index.php/Annotator_User_Guide for NCBO parameters
package ubic.gemma.script
import org.apache.commons.httpclient.HttpClient;
import org.apache.commons.httpclient.methods.PostMethod;
import org.apache.commons.httpclient.params.HttpMethodParams;
import ubic.gemma.loader.entrez.pubmed.*
import ubic.gemma.model.common.description.*

PubMedXMLFetcher f = new PubMedXMLFetcher()
annotatorUrl = "http://rest.bioontology.org/obs/annotator"

// read in pubmed ids
def pm2do = { pmid ->
  try { // test we have a valid id.
    Integer.parseInt(pmid)
  } catch(e) { 
    return
  }

  bibref = f.retrieveByHTTP(Integer.parseInt(pmid))
  text = bibref.title + '\n' + bibref.abstractText + '\n' 
  text += bibref.meshTerms.collect(){ it.term }.join('\n')
  // println(text)
  
  // run through NCBI annotator for DO
  client = new HttpClient()
  client.getParams().setParameter(HttpMethodParams.USER_AGENT,"DOG")

  method = new PostMethod(annotatorUrl)
  method.addParameter("withDefaultStopWords","true"); // default is false
  method.addParameter("ontologiesToExpand", "1009");
  method.addParameter("ontologiesToKeepInResult", "1009"); 
  method.addParameter("isVirtualOntologyId", "true");  // default is false, true is recommended
  method.addParameter("levelMax", "10"); // expand to root, 10 is enough.
  method.addParameter("mappingTypes", "null"); //null, Automatic, Manual 
  method.addParameter("textToAnnotate", text);
  method.addParameter("format", "tabDelimited"); //Options are 'text', 'xml', 'tabDelimited'   
  method.addParameter("apikey", "7b14e900-1ba2-4e58-ac21-c5c4c74e7ece") // Paul's
  
  client.executeMethod(method)
  contents = method.getResponseBodyAsString()
  method.releaseConnection()
  
  // print out pubmed id - DO
  //println(contents)
  contents.eachLine({ println(pmid + '\t' + it) })
}


new File(args[0]).eachLine(pm2do)

// more parameters we are not using
  method.addParameter("longestOnly","false"); // default is false
  method.addParameter("wholeWordOnly","true"); // default is true 
  method.addParameter("filterNumber", "true"); // default is true
  method.addParameter("isStopWordsCaseSensitive","false"); // default is false
  method.addParameter("mintermSize","3"); // default is 3
  method.addParameter("scored", "true"); // default is true
  method.addParameter("withSynonyms","true");  // default is true
  method.addParameter("semanticTypes", "");  // default is ''