cistematic/experiments/genexp.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 import os
  31 import string
  32 from cistematic.programs import supportedPrograms
  33 from cistematic.experiments import experimentTypes, loadExperiment
  34 from os import environ
  35
  36 if environ.get("CISTEMATIC_ROOT"):
  37     cisRoot = environ.get("CISTEMATIC_ROOT")
  38 else:
  39     cisRoot = "/proj/genome"
  40
  41
  42 class GenExp:
  43     dataPath = ""
  44     dbFilePath = ""
  45     expFilePath = ""
  46     expName = ""
  47     expType = ""
  48     dataset = []
  49     expProgs = []
  50     expSettings = {}
  51     analysisName = ""
  52
  53
  54     def __init__(self, path, dbPath, newexp, type="", dset="", progs="", analysis="", createExp=False):
  55         self.dataPath = path
  56         self.dbFilePath = dbPath
  57         self.expName = newexp
  58         self.expFilePath = string.replace(dbPath, ".db", "-db.py")
  59         self.expType = type
  60         self.dataset = dset
  61         self.expProgs = progs
  62         self.analysisName = newexp + analysis
  63
  64         if createExp:
  65             self.createExpDB()
  66
  67         if len(self.expProgs) == 0:
  68             exp = loadExperiment( self.expName , self.dbFilePath)
  69             self.expType = exp.experimentType
  70             self.expProgs = exp.getSetting("expProgs")
  71
  72         if len(dset) == 0:
  73             exp = loadExperiment( self.expName , self.dbFilePath)
  74             self.expType = exp.experimentType
  75             self.dataset = exp.getDataset()
  76
  77         self.createExpRunFile()
  78
  79
  80     def run(self):
  81         exp = loadExperiment( self.expName , self.dbFilePath)
  82         exp.setExternalStatus("QUEUED")
  83         del exp
  84         os.system("cd %s ; nohup /usr/bin/python %s & " % (self.dataPath, self.expFilePath))
  85
  86
  87     def createExpDB():
  88         strexec = 'from cistematic.experiments.' + modName + ' import ' + self.expType + '\n'
  89         exec strexec
  90         strexec = 'exp = ' + self.expType + '("' + self.expName + '", "' + self.dbFilePath + '")\n'
  91         exec strexec
  92
  93
  94     def createExpRunFile(self):
  95         outFile = open(self.expFilePath, "w")
  96         outFile.write(self.expHead())
  97         outFile.write(self.expInit())
  98         outFile.write(self.expPrograms())
  99         outFile.write(self.expRun())
 100         outFile.write(self.expAnalysis())
 101         outFile.write(self.expStatus())
 102         outFile.close()
 103
 104
 105     def expHead(self):
 106         strHead = ["# Experiment: %s autogenerated by Cistematic GenExp 0.9.9b\n" % self.expName]
 107         strHead.append("from sys import path\n")
 108         strHead.append("cisPath = '%s'\n" % cisRoot)
 109         strHead.append("if cisPath not in path:\n\tpath.append(cisPath)\n\n")
 110         for oneProg in supportedPrograms:
 111             strHead.append("from cistematic.programs.%s import %s\n" % (oneProg[0], oneProg[1]))
 112
 113         strHead.append("\n")
 114
 115         return string.join(strHead, "")
 116
 117
 118     def expInit(self):
 119         useMussa = False
 120         useSeqcomp = False
 121         startingGenome = ""
 122         targetGenomes = []
 123         consGene = []
 124         geneIDList = []
 125
 126         if self.expType  in ["orthology", "phyloFoot", "phyloTest"]:
 127             exp = loadExperiment( self.expName , self.dbFilePath)
 128             consSettings = exp.getSetting("expConsConfig")
 129             (window, threshold, numseq, homologs) = consSettings[0].split(":")
 130             self.dataset.reverse()
 131             if homologs != "1" and exp.settingsHasKey("gTargets"):
 132                 gTargets = exp.getSetting("gTargets")
 133                 if len(gTargets) > 0:
 134                     targetGenomes = gTargets[0].split(":")
 135
 136             del exp
 137             for geneID in self.dataset:
 138                 (genome, gid) = geneID
 139                 if (genome not in targetGenomes) and homologs == "1":
 140                     targetGenomes.append(genome)
 141
 142                 if startingGenome == "":
 143                     startingGenome = genome
 144
 145                 gid = gid[0]
 146                 # we are assuming that all of these genes will be from the same
 147                 # genome - this is only true (or necessary) if these are not
 148                 # explicitely maked as "homologs" by the user
 149                 if gid not in consGene:
 150                     consGene.append(gid)
 151
 152                 geneIDList.append((genome, gid))
 153
 154             preInitialize = ""
 155             if homologs == "1":
 156                 preInitializeList = ["exp.createConservation()\n"]
 157                 preInitializeList.append("exp.loadConservation()\n")
 158                 preInitializeList.append("paralogs = %s\n" % str(geneIDList))
 159                 preInitializeList.append('exp.insertHomologs(paralogs, "genExp")\n\n')
 160                 preInitialize = string.join(preInitializeList, "")
 161                 del consGene[1:]
 162
 163             if int(numseq) > 0:
 164                 useSeqcomp = True
 165             else:
 166                 useMussa = True
 167         else:
 168             initializeArguments = str(self.dataset)
 169
 170         strInit = ["# Initialize Experiment\n"]
 171         strInit.append("from cistematic.experiments import loadExperiment\n\n")
 172         strInit.append('exp = loadExperiment("%s", "%s")\n' % (self.expName, self.dbFilePath))
 173         strInit.append('exp.setExternalStatus("INITIALIZING")\n')
 174         strInit.append(preInitialize)
 175         if useMussa or useSeqcomp:
 176             if len(targetGenomes) > 0:
 177                 strInit.append('exp.initialize("%s", %s, %s)\n\n' % (startingGenome, str(consGene), str(targetGenomes)))
 178             else:
 179                 strInit.append('exp.initialize("%s", %s)\n\n' % (startingGenome, str(consGene)))
 180             strInit.append('exp.setExternalStatus("CONSERVATION")\n')
 181             strInit.append("exp.computeAlignments()\n")
 182         else:
 183             strInit.append("exp.initialize(%s)\n\n" % initializeArguments)
 184
 185         if useMussa:
 186             strInit.append("exp.mapMussaConservation(window=%s, threshold=%s)\n\n" % (window, threshold))
 187
 188         if useSeqcomp:
 189             strInit.append("exp.mapSeqcompConservation(window=%s, threshold=%s, minSequences=%s)\n\n" % (window, threshold, numseq))
 190
 191         return string.join(strInit, "")
 192
 193
 194     def expPrograms(self):
 195         index = 1
 196         genomes = []
 197         for (genome, geneID) in self.dataset:
 198             if genome not in genomes:
 199                 genomes.append(genome)
 200
 201         strProg = ["# Load individual programs and settings\n"]
 202         for entry in self.expProgs:
 203             progArray = entry[:-1].split(";")
 204             (amod, aProg) = progArray[0].split(":")
 205             strProg.append("prog%s = %s()\n" % (str(index), aProg))
 206             if len(genomes) == 1 and self.expType == "Simple":
 207                 strProg.apend('prog%s.setGenome("%s")\n' % (str(index), genomes[0]))
 208             if len(progArray) > 1:
 209                 strProg.append("prog%s.setGenExpOptions(%s)\n" % (str(index), str(progArray[1:])))
 210             strProg.append("exp.appendProgram(prog%s)\n" % (str(index)))
 211             index += 1
 212
 213         strProg.append("\n")
 214
 215         return string.join(strProg, "")
 216
 217
 218     def expRun(self):
 219         strRun = ["# Run experiment\n",
 220                   'exp.setExternalStatus("MOTIF FINDING")\n',
 221                   "exp.run()\n",
 222                   "\n"
 223         ]
 224
 225         return string.join(strRun, "")
 226
 227
 228     def expAnalysis(self):
 229         strAnalysis = ["# Analysis Section\n",
 230                        'exp.setExternalStatus("ANALYZING")\n',
 231                        'exp.loadAnalysis("consensus")\n',
 232                        "exp.annotateConsensus()\n",
 233                        "exp.mapConsensus()\n",
 234                        "exp.buildMotifSize()\n",
 235                        "\n",
 236                        'exp.loadAnalysis("1 mismatch")\n',
 237                        "exp.annotateConsensus(numMismatches=1)\n",
 238                        "exp.mapConsensus(numMismatches=1)\n",
 239                        "exp.buildMotifSize()\n",
 240                        "\n",
 241                        'exp.loadAnalysis("PWM - 90% threshold")\n',
 242                        "exp.annotateConsensus()\n",
 243                        "exp.mapMotifs(90.0)\n",
 244                        "exp.buildMotifSize()\n\n",
 245                        "\n"
 246         ]
 247
 248         return string.join(strAnalysis, "")
 249
 250
 251     def expStatus(self):
 252         return "exp.resetExternalStatus()\n\n"