binaries/src/globplot/biopython-1.50/Bio/Search.py

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