annotate bowtie_wrapper.py @ 4:a3825c73805c draft default tip

planemo upload commit 00a7926c285bc4a339bd7deebf40b28f39c7d947-dirty
author devteam
date Tue, 21 Jul 2015 13:03:39 -0400
parents fd0914e451c5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
1 #!/usr/bin/env python
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
2
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
3 """
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
4 Runs Bowtie on single-end or paired-end data.
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
5 For use with Bowtie v. 0.12.7
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
6
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
7 usage: bowtie_wrapper.py [options]
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
8 -t, --threads=t: The number of threads to run
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
9 -o, --output=o: The output file
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
10 --output_unmapped_reads=: File name for unmapped reads (single-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
11 --output_unmapped_reads_l=: File name for unmapped reads (left, paired-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
12 --output_unmapped_reads_r=: File name for unmapped reads (right, paired-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
13 --output_suppressed_reads=: File name for suppressed reads because of max setting (single-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
14 --output_suppressed_reads_l=: File name for suppressed reads because of max setting (left, paired-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
15 --output_suppressed_reads_r=: File name for suppressed reads because of max setting (right, paired-end)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
16 -i, --input1=i: The (forward or single-end) reads file in Sanger FASTQ format
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
17 -I, --input2=I: The reverse reads file in Sanger FASTQ format
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
18 -4, --dataType=4: The type of data (SOLiD or Solexa)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
19 -2, --paired=2: Whether the data is single- or paired-end
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
20 -g, --genomeSource=g: The type of reference provided
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
21 -r, --ref=r: The reference genome to use or index
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
22 -s, --skip=s: Skip the first n reads
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
23 -a, --alignLimit=a: Only align the first n reads
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
24 -T, --trimH=T: Trim n bases from high-quality (left) end of each read before alignment
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
25 -L, --trimL=L: Trim n bases from low-quality (right) end of each read before alignment
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
26 -m, --mismatchSeed=m: Maximum number of mismatches permitted in the seed
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
27 -M, --mismatchQual=M: Maximum permitted total of quality values at mismatched read positions
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
28 -l, --seedLen=l: Seed length
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
29 -n, --rounding=n: Whether or not to round to the nearest 10 and saturating at 30
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
30 -P, --maqSoapAlign=P: Choose MAQ- or SOAP-like alignment policy
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
31 -w, --tryHard=: Whether or not to try as hard as possible to find valid alignments when they exist
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
32 -v, --valAlign=v: Report up to n valid arguments per read
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
33 -V, --allValAligns=V: Whether or not to report all valid alignments per read
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
34 -G, --suppressAlign=G: Suppress all alignments for a read if more than n reportable alignments exist
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
35 -b, --best=b: Whether or not to make Bowtie guarantee that reported singleton alignments are 'best' in terms of stratum and in terms of the quality values at the mismatched positions
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
36 -B, --maxBacktracks=B: Maximum number of backtracks permitted when aligning a read
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
37 -R, --strata=R: Whether or not to report only those alignments that fall in the best stratum if many valid alignments exist and are reportable
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
38 -j, --minInsert=j: Minimum insert size for valid paired-end alignments
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
39 -J, --maxInsert=J: Maximum insert size for valid paired-end alignments
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
40 -O, --mateOrient=O: The upstream/downstream mate orientation for valid paired-end alignment against the forward reference strand
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
41 -A, --maxAlignAttempt=A: Maximum number of attempts Bowtie will make to match an alignment for one mate with an alignment for the opposite mate
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
42 -f, --forwardAlign=f: Whether or not to attempt to align the forward reference strand
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
43 -E, --reverseAlign=E: Whether or not to attempt to align the reverse-complement reference strand
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
44 -F, --offrate=F: Override the offrate of the index to n
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
45 -8, --snpphred=8: SNP penalty on Phred scale
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
46 -6, --snpfrac=6: Fraction of sites expected to be SNP sites
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
47 -7, --keepends=7: Keep extreme-end nucleotides and qualities
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
48 -S, --seed=S: Seed for pseudo-random number generator
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
49 -C, --params=C: Whether to use default or specified parameters
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
50 -u, --iautoB=u: Automatic or specified behavior
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
51 -K, --ipacked=K: Whether or not to use a packed representation for DNA strings
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
52 -Q, --ibmax=Q: Maximum number of suffixes allowed in a block
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
53 -Y, --ibmaxdivn=Y: Maximum number of suffixes allowed in a block as a fraction of the length of the reference
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
54 -D, --idcv=D: The period for the difference-cover sample
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
55 -U, --inodc=U: Whether or not to disable the use of the difference-cover sample
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
56 -y, --inoref=y: Whether or not to build the part of the reference index used only in paired-end alignment
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
57 -z, --ioffrate=z: How many rows get marked during annotation of some or all of the Burrows-Wheeler rows
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
58 -W, --iftab=W: The size of the lookup table used to calculate an initial Burrows-Wheeler range with respect to the first n characters of the query
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
59 -X, --intoa=X: Whether or not to convert Ns in the reference sequence to As
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
60 -N, --iendian=N: Endianness to use when serializing integers to the index file
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
61 -Z, --iseed=Z: Seed for the pseudorandom number generator
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
62 -c, --icutoff=c: Number of first bases of the reference sequence to index
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
63 -x, --indexSettings=x: Whether or not indexing options are to be set
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
64 -H, --suppressHeader=H: Suppress header
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
65 --do_not_build_index: Flag to specify that provided file is already indexed and to just use 'as is'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
66 """
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
67
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
68 import optparse, os, shutil, subprocess, sys, tempfile
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
69
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
70 #Allow more than Sanger encoded variants
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
71 DEFAULT_ASCII_ENCODING = '--phred33-quals'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
72 GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG = { 'fastqsanger':'--phred33-quals', 'fastqillumina':'--phred64-quals', 'fastqsolexa':'--solexa-quals' }
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
73 #FIXME: Integer quality scores are supported only when the '--integer-quals' argument is specified to bowtie; this is not currently able to be set in the tool/wrapper/config
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
74
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
75 def stop_err( msg ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
76 sys.stderr.write( '%s\n' % msg )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
77 sys.exit()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
78
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
79 def __main__():
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
80 #Parse Command Line
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
81 parser = optparse.OptionParser()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
82 parser.add_option( '-t', '--threads', dest='threads', help='The number of threads to run' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
83 parser.add_option( '-o', '--output', dest='output', help='The output file' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
84 parser.add_option( '', '--output_unmapped_reads', dest='output_unmapped_reads', help='File name for unmapped reads (single-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
85 parser.add_option( '', '--output_unmapped_reads_l', dest='output_unmapped_reads_l', help='File name for unmapped reads (left, paired-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
86 parser.add_option( '', '--output_unmapped_reads_r', dest='output_unmapped_reads_r', help='File name for unmapped reads (right, paired-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
87 parser.add_option( '', '--output_suppressed_reads', dest='output_suppressed_reads', help='File name for suppressed reads because of max setting (single-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
88 parser.add_option( '', '--output_suppressed_reads_l', dest='output_suppressed_reads_l', help='File name for suppressed reads because of max setting (left, paired-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
89 parser.add_option( '', '--output_suppressed_reads_r', dest='output_suppressed_reads_r', help='File name for suppressed reads because of max setting (right, paired-end)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
90 parser.add_option( '-4', '--dataType', dest='dataType', help='The type of data (SOLiD or Solexa)' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
91 parser.add_option( '-i', '--input1', dest='input1', help='The (forward or single-end) reads file in Sanger FASTQ format' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
92 parser.add_option( '-I', '--input2', dest='input2', help='The reverse reads file in Sanger FASTQ format' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
93 parser.add_option( '-2', '--paired', dest='paired', help='Whether the data is single- or paired-end' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
94 parser.add_option( '-g', '--genomeSource', dest='genomeSource', help='The type of reference provided' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
95 parser.add_option( '-r', '--ref', dest='ref', help='The reference genome to use or index' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
96 parser.add_option( '-s', '--skip', dest='skip', help='Skip the first n reads' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
97 parser.add_option( '-a', '--alignLimit', dest='alignLimit', help='Only align the first n reads' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
98 parser.add_option( '-T', '--trimH', dest='trimH', help='Trim n bases from high-quality (left) end of each read before alignment' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
99 parser.add_option( '-L', '--trimL', dest='trimL', help='Trim n bases from low-quality (right) end of each read before alignment' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
100 parser.add_option( '-m', '--mismatchSeed', dest='mismatchSeed', help='Maximum number of mismatches permitted in the seed' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
101 parser.add_option( '-M', '--mismatchQual', dest='mismatchQual', help='Maximum permitted total of quality values at mismatched read positions' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
102 parser.add_option( '-l', '--seedLen', dest='seedLen', help='Seed length' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
103 parser.add_option( '-n', '--rounding', dest='rounding', help='Whether or not to round to the nearest 10 and saturating at 30' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
104 parser.add_option( '-P', '--maqSoapAlign', dest='maqSoapAlign', help='Choose MAQ- or SOAP-like alignment policy' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
105 parser.add_option( '-w', '--tryHard', dest='tryHard', help='Whether or not to try as hard as possible to find valid alignments when they exist' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
106 parser.add_option( '-v', '--valAlign', dest='valAlign', help='Report up to n valid arguments per read' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
107 parser.add_option( '-V', '--allValAligns', dest='allValAligns', help='Whether or not to report all valid alignments per read' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
108 parser.add_option( '-G', '--suppressAlign', dest='suppressAlign', help='Suppress all alignments for a read if more than n reportable alignments exist' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
109 parser.add_option( '-b', '--best', dest='best', help="Whether or not to make Bowtie guarantee that reported singleton alignments are 'best' in terms of stratum and in terms of the quality values at the mismatched positions" )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
110 parser.add_option( '-B', '--maxBacktracks', dest='maxBacktracks', help='Maximum number of backtracks permitted when aligning a read' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
111 parser.add_option( '-R', '--strata', dest='strata', help='Whether or not to report only those alignments that fall in the best stratum if many valid alignments exist and are reportable' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
112 parser.add_option( '-j', '--minInsert', dest='minInsert', help='Minimum insert size for valid paired-end alignments' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
113 parser.add_option( '-J', '--maxInsert', dest='maxInsert', help='Maximum insert size for valid paired-end alignments' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
114 parser.add_option( '-O', '--mateOrient', dest='mateOrient', help='The upstream/downstream mate orientation for valid paired-end alignment against the forward reference strand' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
115 parser.add_option( '-A', '--maxAlignAttempt', dest='maxAlignAttempt', help='Maximum number of attempts Bowtie will make to match an alignment for one mate with an alignment for the opposite mate' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
116 parser.add_option( '-f', '--forwardAlign', dest='forwardAlign', help='Whether or not to attempt to align the forward reference strand' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
117 parser.add_option( '-E', '--reverseAlign', dest='reverseAlign', help='Whether or not to attempt to align the reverse-complement reference strand' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
118 parser.add_option( '-F', '--offrate', dest='offrate', help='Override the offrate of the index to n' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
119 parser.add_option( '-S', '--seed', dest='seed', help='Seed for pseudo-random number generator' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
120 parser.add_option( '-8', '--snpphred', dest='snpphred', help='SNP penalty on Phred scale' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
121 parser.add_option( '-6', '--snpfrac', dest='snpfrac', help='Fraction of sites expected to be SNP sites' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
122 parser.add_option( '-7', '--keepends', dest='keepends', help='Keep extreme-end nucleotides and qualities' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
123 parser.add_option( '-C', '--params', dest='params', help='Whether to use default or specified parameters' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
124 parser.add_option( '-u', '--iautoB', dest='iautoB', help='Automatic or specified behavior' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
125 parser.add_option( '-K', '--ipacked', dest='ipacked', help='Whether or not to use a packed representation for DNA strings' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
126 parser.add_option( '-Q', '--ibmax', dest='ibmax', help='Maximum number of suffixes allowed in a block' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
127 parser.add_option( '-Y', '--ibmaxdivn', dest='ibmaxdivn', help='Maximum number of suffixes allowed in a block as a fraction of the length of the reference' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
128 parser.add_option( '-D', '--idcv', dest='idcv', help='The period for the difference-cover sample' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
129 parser.add_option( '-U', '--inodc', dest='inodc', help='Whether or not to disable the use of the difference-cover sample' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
130 parser.add_option( '-y', '--inoref', dest='inoref', help='Whether or not to build the part of the reference index used only in paired-end alignment' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
131 parser.add_option( '-z', '--ioffrate', dest='ioffrate', help='How many rows get marked during annotation of some or all of the Burrows-Wheeler rows' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
132 parser.add_option( '-W', '--iftab', dest='iftab', help='The size of the lookup table used to calculate an initial Burrows-Wheeler range with respect to the first n characters of the query' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
133 parser.add_option( '-X', '--intoa', dest='intoa', help='Whether or not to convert Ns in the reference sequence to As' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
134 parser.add_option( '-N', '--iendian', dest='iendian', help='Endianness to use when serializing integers to the index file' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
135 parser.add_option( '-Z', '--iseed', dest='iseed', help='Seed for the pseudorandom number generator' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
136 parser.add_option( '-c', '--icutoff', dest='icutoff', help='Number of first bases of the reference sequence to index' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
137 parser.add_option( '-x', '--indexSettings', dest='index_settings', help='Whether or not indexing options are to be set' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
138 parser.add_option( '-H', '--suppressHeader', dest='suppressHeader', help='Suppress header' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
139 parser.add_option( '--galaxy_input_format', dest='galaxy_input_format', default="fastqsanger", help='galaxy input format' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
140 parser.add_option( '--do_not_build_index', dest='do_not_build_index', action="store_true", default=False, help='Flag to specify that provided file is already indexed, use as is' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
141 (options, args) = parser.parse_args()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
142 stdout = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
143
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
144 # make temp directory for placement of indices and copy reference file there if necessary
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
145 tmp_index_dir = tempfile.mkdtemp()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
146 # get type of data (solid or solexa)
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
147 if options.dataType == 'solid':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
148 colorspace = '-C'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
149 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
150 colorspace = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
151 # index if necessary
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
152 if options.genomeSource == 'history' and not options.do_not_build_index:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
153 # set up commands
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
154 if options.index_settings =='indexPreSet':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
155 indexing_cmds = '%s' % colorspace
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
156 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
157 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
158 if options.iautoB and options.iautoB == 'set':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
159 iautoB = '--noauto'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
160 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
161 iautoB = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
162 if options. ipacked and options.ipacked == 'packed':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
163 ipacked = '--packed'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
164 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
165 ipacked = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
166 if options.ibmax and int( options.ibmax ) >= 1:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
167 ibmax = '--bmax %s' % options.ibmax
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
168 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
169 ibmax = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
170 if options.ibmaxdivn and int( options.ibmaxdivn ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
171 ibmaxdivn = '--bmaxdivn %s' % options.ibmaxdivn
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
172 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
173 ibmaxdivn = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
174 if options.idcv and int( options.idcv ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
175 idcv = '--dcv %s' % options.idcv
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
176 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
177 idcv = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
178 if options.inodc and options.inodc == 'nodc':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
179 inodc = '--nodc'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
180 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
181 inodc = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
182 if options.inoref and options.inoref == 'noref':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
183 inoref = '--noref'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
184 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
185 inoref = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
186 if options.iftab and int( options.iftab ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
187 iftab = '--ftabchars %s' % options.iftab
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
188 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
189 iftab = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
190 if options.intoa and options.intoa == 'yes':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
191 intoa = '--ntoa'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
192 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
193 intoa = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
194 if options.iendian and options.iendian == 'big':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
195 iendian = '--big'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
196 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
197 iendian = '--little'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
198 if options.iseed and int( options.iseed ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
199 iseed = '--seed %s' % options.iseed
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
200 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
201 iseed = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
202 if options.icutoff and int( options.icutoff ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
203 icutoff = '--cutoff %s' % options.icutoff
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
204 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
205 icutoff = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
206 indexing_cmds = '%s %s %s %s %s %s %s --offrate %s %s %s %s %s %s %s' % \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
207 ( iautoB, ipacked, ibmax, ibmaxdivn, idcv, inodc,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
208 inoref, options.ioffrate, iftab, intoa, iendian,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
209 iseed, icutoff, colorspace )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
210 except ValueError, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
211 # clean up temp dir
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
212 if os.path.exists( tmp_index_dir ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
213 shutil.rmtree( tmp_index_dir )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
214 stop_err( "Something is wrong with the indexing parameters and the indexing and alignment could not be run. Make sure you don't have any non-numeric values where they should be numeric.\n" + str( e ) )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
215 ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
216 ref_file_name = ref_file.name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
217 ref_file.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
218 os.symlink( options.ref, ref_file_name )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
219 cmd1 = 'bowtie-build %s -f %s %s' % ( indexing_cmds, ref_file_name, ref_file_name )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
220 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
221 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
222 tmp_stderr = open( tmp, 'wb' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
223 proc = subprocess.Popen( args=cmd1, shell=True, cwd=tmp_index_dir, stderr=tmp_stderr.fileno() )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
224 returncode = proc.wait()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
225 tmp_stderr.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
226 # get stderr, allowing for case where it's very large
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
227 tmp_stderr = open( tmp, 'rb' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
228 stderr = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
229 buffsize = 1048576
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
230 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
231 while True:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
232 stderr += tmp_stderr.read( buffsize )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
233 if not stderr or len( stderr ) % buffsize != 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
234 break
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
235 except OverflowError:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
236 pass
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
237 tmp_stderr.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
238 if returncode != 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
239 raise Exception, stderr
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
240 except Exception, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
241 # clean up temp dir
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
242 if os.path.exists( tmp_index_dir ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
243 shutil.rmtree( tmp_index_dir )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
244 stop_err( 'Error indexing reference sequence\n' + str( e ) )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
245 stdout += 'File indexed. '
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
246 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
247 ref_file_name = options.ref
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
248 # set up aligning and generate aligning command options
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
249 # automatically set threads in both cases
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
250 tmp_suppressed_file_name = None
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
251 tmp_unmapped_file_name = None
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
252 if options.suppressHeader == 'true':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
253 suppressHeader = '--sam-nohead'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
254 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
255 suppressHeader = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
256 if options.maxInsert and int( options.maxInsert ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
257 maxInsert = '-X %s' % options.maxInsert
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
258 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
259 maxInsert = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
260 if options.mateOrient:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
261 mateOrient = '--%s' % options.mateOrient
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
262 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
263 mateOrient = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
264 quality_score_encoding = GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG.get( options.galaxy_input_format, DEFAULT_ASCII_ENCODING )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
265 if options.params == 'preSet':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
266 aligning_cmds = '-q %s %s -p %s -S %s %s %s ' % \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
267 ( maxInsert, mateOrient, options.threads, suppressHeader, colorspace, quality_score_encoding )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
268 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
269 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
270 if options.skip and int( options.skip ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
271 skip = '-s %s' % options.skip
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
272 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
273 skip = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
274 if options.alignLimit and int( options.alignLimit ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
275 alignLimit = '-u %s' % options.alignLimit
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
276 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
277 alignLimit = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
278 if options.trimH and int( options.trimH ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
279 trimH = '-5 %s' % options.trimH
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
280 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
281 trimH = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
282 if options.trimL and int( options.trimL ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
283 trimL = '-3 %s' % options.trimL
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
284 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
285 trimL = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
286 if options.maqSoapAlign != '-1' and int( options.maqSoapAlign ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
287 maqSoapAlign = '-v %s' % options.maqSoapAlign
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
288 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
289 maqSoapAlign = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
290 if options.mismatchSeed and (options.mismatchSeed == '0' or options.mismatchSeed == '1' \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
291 or options.mismatchSeed == '2' or options.mismatchSeed == '3'):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
292 mismatchSeed = '-n %s' % options.mismatchSeed
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
293 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
294 mismatchSeed = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
295 if options.mismatchQual and int( options.mismatchQual ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
296 mismatchQual = '-e %s' % options.mismatchQual
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
297 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
298 mismatchQual = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
299 if options.seedLen and int( options.seedLen ) >= 5:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
300 seedLen = '-l %s' % options.seedLen
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
301 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
302 seedLen = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
303 if options.rounding == 'noRound':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
304 rounding = '--nomaqround'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
305 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
306 rounding = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
307 if options.minInsert and int( options.minInsert ) > 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
308 minInsert = '-I %s' % options.minInsert
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
309 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
310 minInsert = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
311 if options.maxAlignAttempt and int( options.maxAlignAttempt ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
312 maxAlignAttempt = '--pairtries %s' % options.maxAlignAttempt
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
313 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
314 maxAlignAttempt = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
315 if options.forwardAlign == 'noForward':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
316 forwardAlign = '--nofw'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
317 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
318 forwardAlign = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
319 if options.reverseAlign == 'noReverse':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
320 reverseAlign = '--norc'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
321 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
322 reverseAlign = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
323 if options.maxBacktracks and int( options.maxBacktracks ) > 0 and \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
324 ( options.mismatchSeed == '2' or options.mismatchSeed == '3' ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
325 maxBacktracks = '--maxbts %s' % options.maxBacktracks
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
326 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
327 maxBacktracks = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
328 if options.tryHard == 'doTryHard':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
329 tryHard = '-y'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
330 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
331 tryHard = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
332 if options.valAlign and int( options.valAlign ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
333 valAlign = '-k %s' % options.valAlign
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
334 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
335 valAlign = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
336 if options.allValAligns == 'doAllValAligns':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
337 allValAligns = '-a'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
338 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
339 allValAligns = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
340 if options.suppressAlign and int( options.suppressAlign ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
341 suppressAlign = '-m %s' % options.suppressAlign
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
342 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
343 suppressAlign = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
344 if options.best == 'doBest':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
345 best = '--best'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
346 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
347 best = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
348 if options.strata == 'doStrata':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
349 strata = '--strata'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
350 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
351 strata = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
352 if options.offrate and int( options.offrate ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
353 offrate = '-o %s' % options.offrate
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
354 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
355 offrate = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
356 if options.seed and int( options.seed ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
357 seed = '--seed %s' % options.seed
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
358 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
359 seed = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
360 if options.paired == 'paired':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
361 if options.output_unmapped_reads_l and options.output_unmapped_reads_r:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
362 tmp_unmapped_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir, suffix='.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
363 tmp_unmapped_file_name = tmp_unmapped_file.name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
364 tmp_unmapped_file.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
365 output_unmapped_reads = '--un %s' % tmp_unmapped_file_name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
366 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
367 output_unmapped_reads = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
368 if options.output_suppressed_reads:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
369 tmp_suppressed_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir, suffix='.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
370 tmp_suppressed_file_name = tmp_suppressed_file.name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
371 tmp_suppressed_file.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
372 output_suppressed_reads = '--max %s' % tmp_suppressed_file_name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
373 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
374 output_suppressed_reads = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
375 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
376 if options.output_unmapped_reads:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
377 output_unmapped_reads = '--un %s' % options.output_unmapped_reads
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
378 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
379 output_unmapped_reads = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
380 if options.output_suppressed_reads:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
381 output_suppressed_reads = '--max %s' % options.output_suppressed_reads
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
382 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
383 output_suppressed_reads = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
384 snpfrac = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
385 if options.snpphred and int( options.snpphred ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
386 snpphred = '--snpphred %s' % options.snpphred
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
387 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
388 snpphred = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
389 if options.snpfrac and float( options.snpfrac ) >= 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
390 snpfrac = '--snpfrac %s' % options.snpfrac
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
391 if options.keepends and options.keepends == 'doKeepends':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
392 keepends = '--col-keepends'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
393 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
394 keepends = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
395 aligning_cmds = '-q %s %s -p %s -S %s %s %s %s %s %s %s %s %s %s %s %s ' \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
396 '%s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s ' % \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
397 ( maxInsert, mateOrient, options.threads, suppressHeader,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
398 colorspace, skip, alignLimit, trimH, trimL, maqSoapAlign,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
399 mismatchSeed, mismatchQual, seedLen, rounding, minInsert,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
400 maxAlignAttempt, forwardAlign, reverseAlign, maxBacktracks,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
401 tryHard, valAlign, allValAligns, suppressAlign, best,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
402 strata, offrate, seed, snpphred, snpfrac, keepends,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
403 output_unmapped_reads, output_suppressed_reads,
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
404 quality_score_encoding )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
405 except ValueError, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
406 # clean up temp dir
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
407 if os.path.exists( tmp_index_dir ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
408 shutil.rmtree( tmp_index_dir )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
409 stop_err( 'Something is wrong with the alignment parameters and the alignment could not be run\n' + str( e ) )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
410 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
411 # have to nest try-except in try-finally to handle 2.4
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
412 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
413 # prepare actual mapping commands
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
414 if options.paired == 'paired':
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
415 cmd2 = 'bowtie %s %s -1 %s -2 %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.input2, options.output )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
416 else:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
417 cmd2 = 'bowtie %s %s %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.output )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
418 # align
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
419 tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
420 tmp_stderr = open( tmp, 'wb' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
421 proc = subprocess.Popen( args=cmd2, shell=True, cwd=tmp_index_dir, stderr=tmp_stderr.fileno() )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
422 returncode = proc.wait()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
423 tmp_stderr.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
424 # get stderr, allowing for case where it's very large
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
425 tmp_stderr = open( tmp, 'rb' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
426 stderr = ''
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
427 buffsize = 1048576
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
428 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
429 while True:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
430 stderr += tmp_stderr.read( buffsize )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
431 if not stderr or len( stderr ) % buffsize != 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
432 break
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
433 except OverflowError:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
434 pass
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
435 tmp_stderr.close()
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
436 if returncode != 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
437 raise Exception, stderr
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
438 # get suppressed and unmapped reads output files in place if appropriate
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
439 if options.paired == 'paired' and tmp_suppressed_file_name and \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
440 options.output_suppressed_reads_l and options.output_suppressed_reads_r:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
441 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
442 left = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
443 right = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
444 shutil.move( left, options.output_suppressed_reads_l )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
445 shutil.move( right, options.output_suppressed_reads_r )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
446 except Exception, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
447 sys.stdout.write( 'Error producing the suppressed output file.\n' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
448 if options.paired == 'paired' and tmp_unmapped_file_name and \
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
449 options.output_unmapped_reads_l and options.output_unmapped_reads_r:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
450 try:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
451 left = tmp_unmapped_file_name.replace( '.fastq', '_1.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
452 right = tmp_unmapped_file_name.replace( '.fastq', '_2.fastq' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
453 shutil.move( left, options.output_unmapped_reads_l )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
454 shutil.move( right, options.output_unmapped_reads_r )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
455 except Exception, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
456 sys.stdout.write( 'Error producing the unmapped output file.\n' )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
457 # check that there are results in the output file
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
458 if os.path.getsize( options.output ) == 0:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
459 raise Exception, 'The output file is empty, there may be an error with your input file or settings.'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
460 except Exception, e:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
461 stop_err( 'Error aligning sequence. ' + str( e ) )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
462 finally:
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
463 # clean up temp dir
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
464 if os.path.exists( tmp_index_dir ):
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
465 shutil.rmtree( tmp_index_dir )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
466 stdout += 'Sequence file aligned.\n'
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
467 sys.stdout.write( stdout )
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
468
fd0914e451c5 Uploaded tarball
devteam
parents:
diff changeset
469 if __name__=="__main__": __main__()