Source Code for Module TEES.Utils.InteractionXML.ResolveEPITriggerTypes

  1  import sys, os 
  2  try: 
  3      import xml.etree.cElementTree as ET 
  4  except ImportError: 
  5      import cElementTree as ET 
  6  sys.path.append(os.path.dirname(os.path.abspath(__file__))+"/../../") 
  7  import Utils.ElementTreeUtils as ETUtils 
  8  from collections import defaultdict 
  9   
 10 -def getEPIBaseType(eType): 
 11      if not isNegatableEPITrigger(eType): 
 12          return eType 
 13   
 14      preTag = "" 
 15      if "_" in eType: 
 16          preTag, eType = eType.split("_") 
 17      eType = eType.replace("De", "").replace("de", "") 
 18      if preTag != "": 
 19          return preTag + "_" + eType 
 20      else: 
 21          if eType[0].islower(): 
 22              eType = eType[0].upper() + eType[1:] 
 23          return eType 
 24   
 25 -def negateType(eType): 
 26      preTag = "" 
 27      if "_" in eType: 
 28          preTag, eType = eType.split("_") 
 29      eType = eType.lower() 
 30      if preTag != "": 
 31          return preTag + "_" + "de" + eType 
 32      else: 
 33          return "De" + eType 
 34   
 35 -def isNegatableEPITrigger(eType): 
 36      return eType in ["Phosphorylation", 
 37                       "Dephosphorylation", 
 38                       "Hydroxylation", 
 39                       "Dehydroxylation", 
 40                       "Ubiquitination", 
 41                       "Deubiquitination", 
 42                       "DNA_methylation", 
 43                       "DNA_demethylation", 
 44                       "Glycosylation", 
 45                       "Deglycosylation", 
 46                       "Acetylation", 
 47                       "Deacetylation", 
 48                       "Methylation", 
 49                       "Demethylation"] 
 50                       #"Catalysis"] 
 51   
 52 -def determineNewType(eType, eText): 
 53      assert eText != None 
 54      classNames = eType.split("---") 
 55      newNames = set() 
 56      for className in classNames: 
 57          newNames.add(getNewType(className, eText)) 
 58      return "---".join(sorted(list(newNames))) 
 59       
 60 -def getNewType(eType, eText):     
 61      if not isNegatableEPITrigger(eType): 
 62          return eType 
 63      eBaseType = getEPIBaseType(eType) 
 64      eTextLower = eText.lower() 
 65      if "remov" in eTextLower: 
 66          eNewType = negateType(eBaseType) 
 67      elif "loss" in eTextLower: 
 68          eNewType = negateType(eBaseType) 
 69      elif "erasure" in eTextLower: 
 70          eNewType = negateType(eBaseType) 
 71      #elif eText.startswith("hypo"): 
 72      #    eNewType = negateType(eBaseType) 
 73      elif eTextLower.startswith("de"): 
 74          eNewType = negateType(eBaseType) 
 75      else: 
 76          eNewType = eBaseType 
 77      return eNewType 
 78   
 79 -def negateEvents(input, output=None, verbose=False): 
 80      if not (ET.iselement(input) and input.tag == "sentence"): 
 81          print >> sys.stderr, "Loading corpus file", input 
 82          corpusTree = ETUtils.ETFromObj(input) 
 83          corpusRoot = corpusTree.getroot() 
 84       
 85      if not (ET.iselement(input) and input.tag == "sentence"): 
 86          sentences = corpusRoot.getiterator("sentence") 
 87      else: 
 88          sentences = [input] 
 89      counts = defaultdict(int) 
 90      for sentence in sentences: 
 91          for entity in sentence.findall("entity"): 
 92              counts["all-entities"] += 1 
 93              eType = entity.get("type") 
 94              if not isNegatableEPITrigger(eType): 
 95                  counts["out-of-scope"] += 1 
 96                  continue 
 97              eBaseType = getEPIBaseType(eType) 
 98              eText = entity.get("text").lower() 
 99              eNewType = determineNewType(eType, eText) 
100           
101              # Insert changed charOffset 
102              counts["entities"] += 1 
103              if verbose: 
104                  print "Entity", entity.get("id"), [entity.get("text")], [eType, eBaseType, eNewType], 
105              if eNewType != eBaseType: 
106                  counts["negated"] += 1 
107                  if verbose: print "NEGATED", 
108              if eNewType == eType: 
109                  counts["correct"] += 1 
110                  if verbose: print "CORRECT" 
111              else: 
112                  counts["incorrect"] += 1 
113                  if eNewType == eBaseType: 
114                      counts["incorrect-pos"] += 1 
115                  else: 
116                      counts["incorrect-neg"] += 1 
117                  if verbose: print "INCORRECT" 
118              entity.set("type", eNewType) 
119      if verbose: 
120          print counts 
121       
122      if not (ET.iselement(input) and input.tag == "sentence"): 
123          if output != None: 
124              print >> sys.stderr, "Writing output to", output 
125              ETUtils.write(corpusRoot, output) 
126          return corpusTree                     
127   
128  if __name__=="__main__": 
129      print >> sys.stderr, "##### Extend Triggers #####" 
130      # Import Psyco if available 
131      try: 
132          import psyco 
133          psyco.full() 
134          print >> sys.stderr, "Found Psyco, using" 
135      except ImportError: 
136          print >> sys.stderr, "Psyco not installed" 
137       
138      from optparse import OptionParser 
139      optparser = OptionParser(usage="%prog [options]\nCreate an html visualization for a corpus.") 
140      optparser.add_option("-i", "--input", default=None, dest="input", help="Corpus in analysis format", metavar="FILE") 
141      optparser.add_option("-o", "--output", default=None, dest="output", help="Corpus in analysis format", metavar="FILE") 
142      optparser.add_option("-d", "--debug", default=False, action="store_true", dest="debug", help="") 
143      (options, args) = optparser.parse_args() 
144      assert(options.input != None) 
145      #assert(options.output != None) 
146       
147      negateEvents(options.input, options.output, verbose=options.debug) 
148