Mercurial > repos > dereeper > uniqprimer

comparison uniqprimer-0.5.0/uniqprimer_ok.py @ 3:3249d78ecfc2 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author dereeper
date Mon, 03 Jan 2022 09:56:55 +0000
parents
children
comparison
equal deleted inserted replaced
2:7e0438dad4e9 3:3249d78ecfc2
1 #!/usr/bin/python
2
3 '''
4 Created on Jan 1, 2011
5
6 @author: John L. Herndon
7 @contact: herndon@cs.colostate.edu
8 @organization: Colorado State University
9 @group: Computer Science Department, Asa Ben-Hur's laboratory
10 Mau added the -o primerOutfile -l logfile option
11 '''
12
13 import exceptions
14 import sys
15 import time
16
17 import getopt
18 from primertools import *
19
20 version="0.5.0"
21
22
23 class UniqPrimerFinder( object ):
24
25 def __init__( self, includeFiles, excludeFiles, crossValidate, eprimerOptions):
26
27 utils.logMessage( "UniqPrimerFinder::__init__()", "Initializing UniqPrimerFinder" )
28 self.includeFiles = includeFiles
29 self.includeFileManager = includefilemanager.IncludeFileManager( )
30
31 self.excludeFiles = excludeFiles
32 self.excludeFileManager= excludefilemanager.ExcludeFileManager( )
33
34 self.primerManager = primermanager.PrimerManager( eprimerOptions )
35
36 self.crossValidate = crossValidate
37
38
39 utils.logMessage( "UniqPrimerFinder::__init__()", "Initializing UniqPrimerFinder - complete" )
40
41 def writeOutputFile( self, primers, outputFileName, maxresults = 100 ):
42 '''
43 primers: a list of PrimerSet obs
44 '''
45 ##outputFileName = uPrimer ##Mau: defined this..
46 outputFile = open( outputFileName, 'w' )
47
48 i = 0
49 for primer in primers:
50 i += 1
51
52 outputFile.write( "{0}\t{1}\t{2}\t{3}\n".format( i, primer.forwardPrimer, primer.reversePrimer, primer.productSize ) )
53
54 if i > maxresults:
55 break
56
57 utils.logMessage( "UniqPrimerFinder::writeOutputFile()", "output file written." )
58
59
60 def findPrimers( self, outputFile = "uPrimer.txt" ):
61 outputFile = uPrimer ## Mau adds to overwrite the above value
62
63 utils.logMessage( "UniqPrimerFinder::findPrimers()", "Finding primers for include files" )
64 startTime = time.time( )
65 #generate the combined sequence fasta file for all exclude sequences
66 utils.printProgressMessage( "*** Creating Combined Fasta File for Exclude Files ***" )
67 for excludeFile in self.excludeFiles:
68 self.excludeFileManager.addExcludeFile( excludeFile )
69
70 self.excludeFileManager.exportSequences( )
71
72 self.includeFileManager.setExcludeFile( self.excludeFileManager.getOutputFileName( ) )
73
74 utils.printProgressMessage( "*** Finding Sequences Unique to Target Genome ***" )
75
76 #run nucmer program on all include files
77 for includeFile in self.includeFiles:
78 self.includeFileManager.processIncludeFile( includeFile )
79
80 #get the sequences found in include files, but no the exclude file.
81 uniqueSequences = self.includeFileManager.getUniqueSequences( )
82
83 utils.printProgressMessage( "*** Finding Primers ***" )
84
85 primers = self.primerManager.getPrimers( uniqueSequences )
86
87 if self.crossValidate == True:
88 utils.printProgressMessage( "*** Cross Validating Primers ***" )
89 primers = self.primerManager.crossValidatePrimers( primers, self.excludeFileManager.getOutputFileName( ) )
90
91
92 utils.logMessage( "UniqPrimerFinder::findPrimers( )", "found {0} unique sequences".format( len( primers ) ) )
93
94 self.writeOutputFile( primers, outputFile )
95
96 utils.logMessage( "UniqPrimerFinder::findPrimers()", "Finished finding primers" )
97 endTime = time.time()
98 elapsedMinutes = int( ( endTime - startTime ) / 60 )
99 elapsedSeconds = int( ( endTime - startTime ) % 60 )
100 print "*** Time Elapsed: {0} minutes, {1} seconds ***".format( elapsedMinutes, elapsedSeconds )
101 print "*** Output Written to {0} ***".format( outputFile )
102
103
104 def printUsageAndQuit( ):
105 global version
106 print "uniqprimer - finds primers unique to a genome"
107 print "Version: " + str( version )
108 print "Summary of Options."
109 print "Required Arguments:"
110 print " -i <filename>: use <filename> as an include file. Primers will be identified for this genome"
111 print " -x <filename>: use <filename> as an exclude file. Primers for this genome will be excluded"
112 print " -o <filename>: specify the name of the unique primer output file (default is uPrimer.txt)" ## Mau added..
113 print " -l <filename>: specify the name of the log output file" ## Mau added..
114
115 print "\nOptional Arguments:"
116 print " --productsizerage: set a range for the desired size of PCR product (default=200-250). Example: ./uniqprimer -productsizerage 100-150"
117 print " --primersize: set the desired primer size (default=20)"
118 print " --minprimersize: set the minimum primer size (default=27)"
119 print " --maxprimersize: set the maximum primer size (default=18)"
120 print " --crossvalidate: force the program to cross validate primers against exclude files for extra certainty"
121 print " --keeptempfiles: force the program to keep temporary files"
122
123 print "\n\nExample:"
124 print "uniqprimer -i <includefile1> -i <includefile2> ... -i <includefileN> -x <excludefile1> -x <excludefile2> ... -x <excludefileN> -o primers.txt -l logfile.txt"
125 utils.shutdownLogging( )
126 sys.exit( )
127
128
129 opts = 'i:x:h:o:l:' # Mau added :o & :l for outfile specification
130 longopts=[ "productsizerange=", "primersize=", "minprimersize=", "maxprimersize=", "crossvalidate", "keeptempfiles" ]
131
132 def parseArgs( args ):
133
134
135 global uPrimer ## Mau added lf, brute force...
136 global lf # Mau added lf, brute force...
137 #uPrimer = "uPrimer.txt" ##the default value...
138
139 crossValidate = False
140 cleanup = True
141 optlist, args = getopt.getopt( args, opts, longopts )
142
143 includeFiles = [ ]
144 excludeFiles = [ ]
145 eprimerOptions = utils.EPrimerOptions( )
146
147 verbose = False
148 for opt in optlist:
149 if opt[ 0 ] == '-i':
150 includeFiles.append( opt[ 1 ] )
151 elif opt[ 0 ] == '-x':
152 excludeFiles.append( opt[ 1] )
153 elif opt[ 0 ] == '-v':
154 verbose = True
155 elif opt[ 0 ] == '-o': ## Mau added, if -o...
156 uPrimer = str(opt[1]) ## Mau added, then get filename for outfile after -o
157 elif opt[ 0 ] == '-l': ## Mau added, if -l...
158 lf = str(opt[1]) ## Mau added, then get filename for logfile after -l
159 elif opt[ 0 ] == '--productsizerange':
160 eprimerOptions.setProductRange( opt[ 1 ] )
161 productsizerange = opt[ 1 ]
162 elif opt[ 0 ] == '--primersize':
163 eprimerOptions.setPrimerSize( opt[1 ] )
164 elif opt[ 0 ] == '--minprimersize':
165 eprimerOptions.setMinPrimerSize( opt[1 ] )
166 elif opt[ 0 ] == '--maxprimersize':
167 eprimerOptions.setMaxPrimerSize( opt[1 ] )
168 elif opt[ 0 ] == '--crossvalidate':
169 crossValidate = True
170 elif opt[ 0 ] == '--crossvalidate':
171 crossValidate = True
172 elif opt[ 0 ] == '--keeptempfiles':
173 cleanup = False
174 elif opt[ 0 ] == '-h':
175 printUsageAndQuit( )
176 else:
177 print "Unknown option: " + str( opt[ 0 ] )
178 printUsageAndQuit( )
179 #print "uPrimer: " + uPrimer + " log file name: " + lf + "\n"
180 if len( includeFiles ) == 0 or len( excludeFiles ) == 0:
181
182 print "You must specify at least one include file and at least one exclude file"
183 printUsageAndQuit( )
184
185 return includeFiles, excludeFiles, crossValidate, cleanup, verbose, eprimerOptions, lf , uPrimer #Mau: add lf, uPrime
186
187 def main( args, debug = False):
188 #parse the command line arguments for include and exclude files
189
190 includeFiles, excludeFiles, crossValidate, cleanup, verbose, eprimerOptions, lf, uPrimer = parseArgs( args ) ##Mau add: lf
191
192 utils.initialize( True, cleanup, lf) ##Mau: add lf
193
194 #find primers for the include sequences
195
196 try:
197 utils.logMessage( "uniqprimer::Main( )", "Logging include files: " )
198 utils.logList( "uniqprimer::Main( )", includeFiles )
199 utils.logMessage( "uniqprimer::Main( )", "Logging exclude files: " )
200 utils.logList( "uniqprimer::Main( )", excludeFiles)
201 print "*** Finding Primers ***"
202 uniqPrimer = UniqPrimerFinder( includeFiles, excludeFiles, crossValidate, eprimerOptions)
203 uniqPrimer.findPrimers( )
204 except utils.NoFileFoundException as nfe:
205 print "File not found: " + str( nfe.filename )
206 printUsageAndQuit( )
207 except utils.ProgramNotFoundException as pnfe:
208 print str( pnfe.programName ) + ": program is not installed or is not in your path."
209 print str( pnfe.details )
210 except utils.NoPrimersExistException as npe:
211 print "Failure: No unique primers exist for this combination"
212 except exceptions.BaseException as e:
213 print "It appears that an unknown sequence of events has resulted in the internal explosion of this program. Please send the file called \'log_uniqprimer.txt\' to herndon@cs.colostate.edu and tell that bonehead John to fix it!"
214 print "Details:"
215 print e
216
217 utils.shutdown( )
218
219 print "*** Finished ***"
220
221 if __name__ == '__main__':
222
223 #temp_args = "-i data/testdata/smallinclude.ffn -x data/testdata/smallexclude.ffn".split( )
224
225 #temp_args = "-i data/XOO_MAI1_scaffolds.fas -x data/KACC.ffn".split( )
226 if len( sys.argv ) == 1:
227 printUsageAndQuit( )
228 main( sys.argv[ 1: ], debug = True )
229
230
231
232
233
234
235
236
237
238
239
240
241