Source Code for Module Bio.builders.SeqRecord.sequence

  1  from xml.sax import handler 
  2   
  3  from Martel import Dispatch 
  4   
  5  from Bio import SeqRecord, StdHandler, Std, DBXRef, Seq 
  6  from Bio import Alphabet, SeqFeature 
  7  from Bio.Alphabet import IUPAC 
  8   
  9  alphabet_table = { 
 10      "iupac-protein": IUPAC.protein, 
 11      "iupac-dna": IUPAC.unambiguous_dna, 
 12      "iupac-rna": IUPAC.unambiguous_rna, 
 13      "iupac-extended-protein": IUPAC.extended_protein, 
 14      "iupac-ambiguous-dna": IUPAC.ambiguous_dna, 
 15      "iupac-ambiguous-rna": IUPAC.ambiguous_rna, 
 16      "protein": Alphabet.generic_protein, 
 17      "dna": Alphabet.generic_dna, 
 18      "rna": Alphabet.generic_rna, 
 19      "unknown": Alphabet.single_letter_alphabet, 
 20      } 
 21   
 22   
 23  # Convert from the internal Feature data structure used by the parser 
 24  # into the standard Biopytho form 
 25 -def convert_std_feature(feature): 
 26      f = SeqFeature.SeqFeature() 
 27      loc = feature.location 
 28      return feature 
 29   
 30 -class BuildSeqRecord(Dispatch.Dispatcher): 
 31 -    def __init__(self): 
 32          Dispatch.Dispatcher.__init__(self) 
 33          self.acquire(StdHandler.Handle_dbid(self.add_dbid), 
 34                       prefix = Std.NS) 
 35          self.acquire(StdHandler.Handle_dbxref(self.add_dbxref_dbids), 
 36                       prefix = Std.NS) 
 37          self.acquire(StdHandler.Handle_description(self.add_description), 
 38                       prefix = Std.NS) 
 39          self.acquire(StdHandler.Handle_sequence(self.add_sequence), 
 40                       prefix = Std.NS) 
 41          self.acquire(StdHandler.Handle_features(self.add_features), 
 42                       prefix = Std.NS) 
 43          self.acquire(StdHandler.Handle_dbxref(self.add_dbxref), 
 44                       prefix = Std.NS) 
 45   
 46 -    def start_record(self, tag, attrs): 
 47          self.dbname = None 
 48          self.id_text = None 
 49          self.name_text = '<unknown name>' 
 50          self.description = None 
 51          self.alphabet = None 
 52          self.seq = None 
 53          self.features = None 
 54          self.dbxrefs = [] 
 55   
 56 -    def add_dbid(self, text, attrs): 
 57          if attrs.get("type") == "primary": 
 58              self.dbname = attrs.get("dbname", "unknown") 
 59              self.id_text = text 
 60          # use the first accession/secondary id as the name 
 61          # this should be equivalent to what Biopython does 
 62          elif attrs.get("type") in ["accession", "secondary"]: 
 63              self.name_text = text 
 64   
 65 -    def add_dbxref_dbids(self, dbname_style, dbname, idtype, dbid, negate): 
 66          """Handle setting name and id attributes from the dbxref ids. 
 67   
 68          Likely we'll either have a dbid or dbxref dbids to use. We default 
 69          to using the dbid if it exists. 
 70          """ 
 71          # first deal with the primary id: SeqFeature.id 
 72          # set the id if we haven't yet set an id (take the first id we get) 
 73          # and if we have a primary id 
 74          if (self.id_text is None and idtype == "primary"): 
 75              self.id_text = dbid 
 76   
 77          # now deal with secondary ids: SeqFeature.name 
 78          if idtype == "secondary": 
 79              self.name_text = dbid 
 80   
 81   
 82 -    def add_description(self, text): 
 83          self.description = text 
 84   
 85 -    def add_sequence(self, (alphabet, seq, gapchar, stopchar)): 
 86          alphabet = alphabet_table.get(alphabet, 
 87                                        Alphabet.single_letter_alphabet) 
 88          self.seq = Seq.Seq(seq, alphabet) 
 89   
 90 -    def add_dbxref(self, dbname_style, dbname, idtype, dbid, negate): 
 91          """Store all id cross references. 
 92          """ 
 93          self.dbxrefs.append(DBXRef.from_parser(dbname_style, dbname, idtype, 
 94                                                 dbid, negate)) 
 95   
 96 -    def add_features(self, features): 
 97          # Brad -- I can't understand this assertion -- there is no 
 98          # self.features on the first call to add features and then you'll 
 99          # expect to have some on future calls 
100          # assert self.features is None 
101          #print [feature.location for feature in features] 
102          self.features = map(convert_std_feature, features) 
103   
104 -    def end_record(self, tag): 
105          self.document = SeqRecord.SeqRecord( 
106              seq = self.seq, 
107              id = self.id_text, 
108              name = self.name_text, 
109              description = self.description, 
110              dbxrefs = self.dbxrefs, 
111              features = self.features, 
112              ) 
113           
114  make_builder = BuildSeqRecord 
115