cistematic/genomes/spurpuratus.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 # data for Strongylocentrotus purpuratus
  31 import string
  32 from cistematic.genomes import Genome
  33 from os import environ
  34
  35 if environ.get("CISTEMATIC_ROOT"):
  36     cisRoot = environ.get("CISTEMATIC_ROOT")
  37 else:
  38     cisRoot = "/proj/genome"
  39
  40 geneDB = "%s/S_purpuratus/spurpuratus.genedb" % cisRoot
  41
  42
  43 def loadChromosome(db, chromPath, chromOutPath):
  44     seqArray = []
  45     seqLen = 0
  46     spGenome = Genome("spurpuratus", dbFile=db)
  47     inFile = open(chromPath, "r")
  48     header = inFile.readline()
  49     while header != "":
  50         seqArray = []
  51         seqLen = 0
  52         fields = header.split()
  53         if "purpuratus" in header:
  54             idpart = fields[3]
  55         else:
  56             idpart = fields[-1].strip()
  57
  58         parts = idpart.split("_")
  59         chromID = parts[-1]
  60         currentLine = inFile.readline()
  61         while currentLine != "" and currentLine[0] != ">":
  62             lineSeq = currentLine.strip()
  63             seqLen += len(lineSeq)
  64             seqArray.append(lineSeq)
  65             currentLine = inFile.readline()
  66
  67         seq = string.join(seqArray, "")
  68         if seqLen < 50000:
  69             print "Added contig %s to database" % chromID
  70             spGenome.addSequence(("spurpuratus", chromID), seq, "chromosome", str(seqLen))
  71             spGenome.addChromosomeEntry(chromID, chromID, "db")
  72         else:
  73             outFileName = "%s%s.bin" % (chromOutPath, chromID)
  74             outFile = open("%s%s" % (cisRoot, outFileName), "w")
  75             outFile.write(seq)
  76             outFile.close()
  77             print "Added contig file %s to database" % outFileName
  78             spGenome.addChromosomeEntry(chromID, outFileName, "file")
  79
  80         header = currentLine
  81
  82     inFile.close()
  83
  84
  85 def createDBFile(db):
  86     spGenome = Genome("spurpuratus", version="2.1",  dbFile=db)
  87     spGenome.createGeneDB(db)
  88
  89
  90 def createDBindices(db):
  91     spGenome = Genome("spurpuratus", version="2.1", dbFile=db)
  92     spGenome.createIndices()
  93
  94
  95 def buildSpurpuratusDB(db=geneDB):
  96     chromoPath = "%s/download/Spur2.1_Nmasked.txt" % cisRoot
  97     chromoOutPath = "/S_purpuratus/"
  98
  99     print "Creating database %s" % db
 100     createDBFile(db)
 101
 102     print "Loading genomic sequence"
 103     loadChromosome(db, chromoPath, chromoOutPath)
 104
 105     print "Creating Indices"
 106     createDBindices(db)
 107
 108     print "Finished creating database %s" % db