Source Code for Module TEES.ExampleBuilders.FeatureBuilders.RELFeatureBuilder

  1  from FeatureBuilder import FeatureBuilder 
  2   
  3  # Amino acids from http://www.bio.davidson.edu/courses/genomics/jmol/aatable.html 
  4  #amino acid     three letter code     single letter code 
  5   
  6  subcomponent = set(["region", "promoter", "upstream", "fragment", "site", 
  7                "sequence", "segment", "repeat", "repeat", "element", 
  8                "duplication", "exon", "downstream", "terminus", "motif", 
  9                "frame", "carboxy-terminus", "domain", "subunit", "codon", 
 10                "promoter", "enhancer", "locus", "ltr", "helix-loop-helix", 
 11                "zinc-finger", "portion", "residue", "box", "intron"]) 
 12   
 13  supergroup = set(["complex", "family", "octamer", "microtubule"]) 
 14   
 15  aminoAcids = [ 
 16      #nonpolar (hydrophobic) 
 17      ("glycine", "gly", "g", "nonpolar", "neutral"),  
 18      ("alanine", "ala", "a", "nonpolar", "neutral"), 
 19      ("valine", "val", "v", "nonpolar", "neutral"), 
 20      ("leucine", "leu", "l", "nonpolar", "neutral"), 
 21      ("isoleucine", "ile", "i", "nonpolar", "neutral"), 
 22      ("methionine", "met", "m", "nonpolar", "neutral"), 
 23      ("phenylalanine", "phe", "f", "nonpolar", "neutral"), 
 24      ("tryptophan", "trp", "w", "nonpolar", "neutral"), 
 25      ("proline", "pro", "p", "nonpolar", "neutral"),  
 26      #polar (hydrophilic) 
 27      ("serine", "ser", "s", "hydrophilic", "neutral"), 
 28      ("threonine", "thr", "t", "hydrophilic", "neutral"), 
 29      ("cysteine", "cys", "c", "hydrophilic", "neutral"), 
 30      ("tyrosine", "tyr", "y", "hydrophilic", "neutral"), 
 31      ("asparagine", "asn", "n", "hydrophilic", "neutral"), 
 32      ("glutamine", "gln", "q", "hydrophilic", "neutral"), 
 33      #electrically charged (negative and hydrophilic) 
 34      ("aspartic acid", "asp", "d", "hydrophilic", "negative"), 
 35      ("glutamic acid", "glu", "e", "hydrophilic", "negative"), 
 36      #electrically charged (positive and hydrophilic) 
 37      ("lysine", "lys", "k", "hydrophilic", "positive"), 
 38      ("arginine", "arg", "r", "hydrophilic", "positive"), 
 39      ("histidine", "his", "h", "hydrophilic", "positive")] 
 40   
 41 -class RELFeatureBuilder(FeatureBuilder): 
 42 -    def __init__(self, featureSet): 
 43          FeatureBuilder.__init__(self, featureSet) 
 44          #self.noAnnType = False 
 45          #self.edgeTypesForFeatures = [] 
 46          #self.useNonNameEntities = False 
 47   
 48 -    def findAminoAcid(self, string): 
 49          global aminoAcids 
 50       
 51          string = string.lower() 
 52          for aa in aminoAcids: 
 53              word = string.find(aa[0]) 
 54              if word != -1: 
 55                  return word, aa 
 56              else: 
 57                  tlc = string.find(aa[1]) # three letter code 
 58                  if tlc != -1: 
 59                      # Three letter code must not be a part of a word (where it could be just a substring) 
 60                      if (tlc == 0 or not string[tlc-1].isalpha()) and (tlc + 3 >= len(string) or not string[tlc + 3].isalpha()): 
 61                          return tlc, aa 
 62          return -1, None 
 63       
 64 -    def buildAllFeatures(self, tokens, tokenIndex): 
 65          token = tokens[tokenIndex] 
 66          tokText = token.get("text").lower() 
 67           
 68          self.buildAminoAcidFeatures(tokText) 
 69          self.buildDNAFeatures(tokText) 
 70          self.buildSubstringFeatures(tokens, tokenIndex) 
 71          self.buildRangeFeatures(tokens, tokenIndex) 
 72          self.buildKnownWordFeatures(tokText) 
 73       
 74 -    def buildAminoAcidFeatures(self, string): 
 75          index, aa = self.findAminoAcid(string) 
 76          if aa != None: 
 77              self.setFeature("RELaminoacid_string") 
 78              self.setFeature("RELaminoacid_" + aa[1]) 
 79       
 80 -    def findSubstring(self, string, substring, tag=None): 
 81          if tag == None: 
 82              tag = substring 
 83          index = string.find(substring) 
 84          if index != -1: 
 85              self.setFeature("RELsubstring_"+tag) 
 86              if index + len(substring) == len(string): 
 87                  self.setFeature("RELsubstring_terminal_"+tag) 
 88              else: 
 89                  self.setFeature("RELsubstring_nonterminal_"+tag) 
 90               
 91 -    def buildSubstringFeatures(self, tokens, tokenIndex): 
 92          string = "" 
 93          for t in tokens[tokenIndex-6:tokenIndex]: 
 94              # TODO the actual token does not seem to be included 
 95              string += t.get("text") 
 96          string = string.lower().replace("-", "").replace(" ", "") 
 97          # nfkb 
 98          self.findSubstring(string, "nfkappab", "nfkb") 
 99          self.findSubstring(string, "nfkb") 
100          self.findSubstring(string, "nfkappab", "complex") 
101          self.findSubstring(string, "nfkb", "complex") 
102          # kappa-b 
103          self.findSubstring(string, "kappab") 
104          # ap-1 
105          self.findSubstring(string, "ap1") 
106          self.findSubstring(string, "activatingprotein1", "ap1") 
107          self.findSubstring(string, "ap1", "complex") 
108          self.findSubstring(string, "activatingprotein1", "complex") 
109          # proteasome 
110          self.findSubstring(string, "proteasome") 
111          self.findSubstring(string, "proteasome", "complex") 
112          # base pairs 
113          self.findSubstring(string, "bp", "bp") 
114          self.findSubstring(string, "basepair", "bp") 
115          # primes 
116          self.findSubstring(string, "5'", "5prime") 
117          self.findSubstring(string, "3'", "3prime") 
118       
119 -    def buildDNAFeatures(self, string): 
120          for letter in string: 
121              if letter not in ["a", "g", "t", "c"]: 
122                  return 
123          self.setFeature("RELDNA_sequence") 
124       
125 -    def buildRangeFeatures(self, tokens, tokenIndex): 
126          if tokenIndex > 1: 
127              if tokens[tokenIndex-1].get("text").lower() in ["to", "and", "-"]: 
128                  t1Text = tokens[tokenIndex-2].get("text") 
129                  if t1Text[0] == "-" or t1Text[0] == "+": 
130                      t1Text = t1Text[1:] 
131                  t2Text = tokens[tokenIndex].get("text") 
132                  if t2Text[0] == "-" or t2Text[0] == "+": 
133                      t2Text = t2Text[1:] 
134                  if t1Text.isdigit() and t2Text.isdigit(): 
135                      self.setFeature("RELnumeric_range") 
136       
137 -    def buildKnownWordFeatures(self, string): 
138          global subcomponent, supergroup 
139           
140          string = string.lower() 
141           
142          if string[-1] == "s": 
143              singular = string[:-1] 
144          else: 
145              singular = None 
146          if string in subcomponent or singular in subcomponent: 
147              self.setFeature("RELknown_subcomponent") 
148          if string in supergroup or singular in supergroup: 
149              self.setFeature("RELknown_supergroup") 
150