Source Code for Module Bio.Search

  1  import warnings 
  2  from Bio import BiopythonDeprecationWarning 
  3  warnings.warn("Long obsolete module Bio/Search.py is deprecated.", 
  4                BiopythonDeprecationWarning) 
  5   
  6   
  7  # BLASTN 2.0a19MP-WashU [05-Feb-1998] [Build decunix3.2 01:53:29 05-Feb-1998] 
  8  # BLASTP 2.0.4 [Feb-24-1998] 
  9 -class Algorithm(object): 
 10 -    def __init__(self, name, version, description = ""): 
 11          self.name = name                 # 'blastx', 'blastn', etc. 
 12          self.version = version           # '2.1.2' or '2.0a19MP-WashU' 
 13          self.description = description   # '[05-Feb-1998] [Build dec ...1998]' 
 14   
 15   
 16  # Query=  YAL001C YAL001C, Chr I from 147596 to 147665, and 147756 to 151168, 
 17  #     reverse complement 
 18  #         (3483 letters) 
 19 -class Query(object): 
 20 -    def __init__(self, name, accession, description, length): 
 21          self.name = name                # 'YAL001C' 
 22          self.accession = accession      # or None if missing 
 23          self.description = description  # 'YAL001C, Chr I from 147596 to ... ' 
 24          self.length = length            # 3483 
 25   
 26   
 27  # Database:  ArabidopsisN 
 28  #            66,211 sequences; 69,074,155 total letters. 
 29 -class Database(object): 
 30 -    def __init__(self, name, letters, entries): 
 31          self.name = name            # ArabidopsisN 
 32          self.letters = letters      # 69074155 
 33          self.entries = entries      # 66211 
 34   
 35   
 36 -class TableInfo(object): 
 37 -    def __init__(self, full_description, info): 
 38          self.__dict__.update(info) 
 39          self.full_description = full_description 
 40   
 41   
 42 -class Search(object): 
 43 -    def __init__(self, algorithm, query, database, table, hits, 
 44                   parameters, statistics): 
 45          self.algorithm = algorithm 
 46          self.query = query 
 47          self.database = database 
 48          self.table = table 
 49          self.hits = hits 
 50          self.parameters = parameters 
 51          self.statistics = statistics 
 52   
 53   
 54 -class Hit(object): 
 55 -    def __init__(self, name, description, accession, length, 
 56                   algorithm, hsps = None): 
 57          self.name = name 
 58          self.description = description 
 59          self.accession = accession 
 60          self.length = length 
 61          self.algorithm = algorithm 
 62          if hsps is None: 
 63              hsps = [] 
 64          self.hsps = hsps 
 65   
 66 -    def __len__(self): 
 67          return self.length 
 68   
 69   
 70  # >GB_PL:ATF18F4 AL021637 Arabidopsis thaliana DNA chromosome 4, BAC clone 
 71  #           F18F4 (ESSAII project). 2/98 
 72  #             Length = 93,646 
 73  # 
 74  #   Minus Strand HSPs: 
 75  # 
 76  #  Score = 226 (33.9 bits), Expect = 0.80, P = 0.55 
 77  #  Identities = 98/142 (69%), Positives = 98/142 (69%), Strand = Minus / Plus 
 78  #    [...lines deleted...] 
 79  # Query:  2486 ATATCAAGCAATTTGATAAGATCTAG 2461 
 80  #              A AT  A C ATT GA AAGATC AG 
 81  # Sbjct: 85387 AGATTTACCTATT-GAGAAGATCAAG 85411 
 82   
 83  # computed from the strings 
 84 -class _SeqLength: 
 85 -    def __init__(self, length, identical, positives, gaps): 
 86          self.length = length 
 87          self.identical = identical 
 88          self.positives = positives 
 89          self.gaps = gaps 
 90   
 91 -    def __len__(self): 
 92          return self.length 
 93   
 94 -    def __getattr__(self, name): 
 95          if name == "frac_identical": 
 96              return float(self.identical) / self.length 
 97          elif name == "frac_positives": 
 98              return float(self.positives) / self.length 
 99          raise AttributeError(name) 
100   
101   
102 -class HomologySeq(_SeqLength): 
103 -    def __init__(self, seq, identical, positives, gaps): 
104          _SeqLength.__init__(self, len(seq), identical, positives, gaps) 
105          self.seq = seq 
106   
107   
108 -class HSPSeq(_SeqLength): 
109 -    def __init__(self, name, seq, location, identical, positives, gaps): 
110          _SeqLength.__init__(self, len(seq), identical, positives, gaps) 
111          self.name = name 
112          self.seq = seq 
113          self.location = location 
114   
115   
116 -class HSP(_SeqLength): 
117 -    def __init__(self, 
118                   query_seq,    # ATATCAAGCAATTTGATAAGATCTAG 
119                   homology_seq, # A AT  A C ATT GA AAGATC AG 
120                   subject_seq,  # AGATTTACCTATT-GAGAAGATCAAG 
121   
122                   query_location,   # (2486, 2461, negative strand) 
123                   subject_location, # (85387, 85411) 
124   
125                   query_name,     # Query (or None) 
126                   subject_name,   # Sbjct (or None) 
127   
128                   algorithm,  # an Algorithm 
129                   info,       # contains Key/value pairs 
130                   homology_gaps = None,  # Is this needed? 
131                   ): 
132          assert len(query_seq) == len(homology_seq) == len(subject_seq), \ 
133                 (query_seq, homology_seq, subject_seq) 
134          self.algorithm = algorithm 
135   
136          query_gaps = query_seq.count("-") 
137          subject_gaps = subject_seq.count("-") 
138          if homology_gaps is None: 
139              homology_gaps = query_gaps + subject_gaps 
140          self.info = info 
141   
142          identical = info["identical"] 
143          # bioperl calls this 'conserved' 
144          positives = info.get("positives", identical) 
145   
146          _SeqLength.__init__(self, len(query_seq), identical, 
147                              positives, homology_gaps) 
148   
149          self.query = HSPSeq(name = query_name, 
150                              seq = query_seq, 
151                              location = query_location, 
152                              identical = identical, 
153                              positives = positives, 
154                              gaps = query_gaps) 
155   
156          self.subject = HSPSeq(name = subject_name, 
157                                seq = subject_seq, 
158                                location = subject_location, 
159                                identical = identical, 
160                                positives = positives, 
161                                gaps = subject_gaps) 
162          self.homology = HomologySeq(seq = homology_seq, 
163                                      identical = identical, 
164                                      positives = positives, 
165                                      gaps = homology_gaps) 
166