cistematic/experiments/fasta.py

   1 ###########################################################################
   2 #                                                                         #
   3 # C O P Y R I G H T   N O T I C E                                         #
   4 #  Copyright (c) 2003-10 by:                                              #
   5 #    * California Institute of Technology                                 #
   6 #                                                                         #
   7 #    All Rights Reserved.                                                 #
   8 #                                                                         #
   9 # Permission is hereby granted, free of charge, to any person             #
  10 # obtaining a copy of this software and associated documentation files    #
  11 # (the "Software"), to deal in the Software without restriction,          #
  12 # including without limitation the rights to use, copy, modify, merge,    #
  13 # publish, distribute, sublicense, and/or sell copies of the Software,    #
  14 # and to permit persons to whom the Software is furnished to do so,       #
  15 # subject to the following conditions:                                    #
  16 #                                                                         #
  17 # The above copyright notice and this permission notice shall be          #
  18 # included in all copies or substantial portions of the Software.         #
  19 #                                                                         #
  20 # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,         #
  21 # EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF      #
  22 # MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND                   #
  23 # NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS     #
  24 # BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN      #
  25 # ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN       #
  26 # CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE        #
  27 # SOFTWARE.                                                               #
  28 ###########################################################################
  29 #
  30 # a simple use of the experiment class
  31 from experiment import Experiment
  32 from draw import Draw
  33 from analyzeMotifs import AnalyzeMotifs
  34 import sys
  35
  36 class Fasta(Experiment, AnalyzeMotifs, Draw):
  37     experimentType = "fasta"
  38
  39
  40     def run(self, fastaFile):
  41         self.loadFasta(fastaFile)
  42         Experiment.run(self)
  43         datasetID = self.genepoolID
  44         for (prog, settingsID) in self.programs:
  45             if 1:
  46                 prog.inputFile(fastaFile)
  47                 settings = self.getSettingsID(settingsID)[1]
  48                 prog.setSettings(eval(settings))
  49                 runID = self.setRun(prog.name(), datasetID, settingsID)
  50                 tag = str(runID)
  51                 prog.setTagID(tag)
  52                 prog.run()
  53                 theMotifs = prog.getMotifs()
  54                 for mot in theMotifs:
  55                     self.appendResults(mot)
  56             else:
  57                 self.mlog("Error running program %s with settings %s" % (prog, settingsID))
  58
  59
  60     def loadFasta(self, ffile):
  61         """ load fasta file into genepool
  62         """
  63         f=open(ffile, "r")
  64         line = f.readline()
  65         i = 1
  66         while line != "":
  67             seq = ""
  68             templine = f.readline()
  69             while templine != "" and templine[0] != ">":
  70                 seq = seq + templine[0:-1]
  71                 templine = f.readline()
  72                 name = "seq%s" % line.strip()[1:]
  73                 # "progress bar" of dots...
  74                 if (i % 10 == 0):
  75                     sys.stderr.write(".")
  76
  77                 if (i % 1000 ==  0):
  78                     sys.stderr.write("%s\n" % i)
  79                 else:
  80                     if (i % 100 == 0):
  81                         sys.stderr.write(" ")
  82
  83                 i = i + 1
  84
  85             self.genepool[(ffile, name)] = seq
  86             line = templine
  87             print
  88         f.close()