annotate ssake.py @ 0:0ec408bcfc80 draft

Uploaded
author crs4
date Wed, 11 Sep 2013 12:51:21 -0400
parents
children 386166019772
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
1 # -*- coding: utf-8 -*-
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
2 """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
3 SSAKE wrapper
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
4 """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
5
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
6 import logging
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
7 import optparse
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
8 import os
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
9 import shutil
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
10 import subprocess
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
11 import tempfile
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
12
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
13 def execute(cmd):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
14 """ """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
15 subprocess.check_call(args=cmd, stdout=open(os.devnull, 'w'), shell=True)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
16
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
17
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
18 def which(name, flags=os.X_OK):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
19 """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
20 Search PATH for executable files with the given name.
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
21 """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
22 result = []
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
23 exts = filter(None, os.environ.get('PATHEXT', '').split(os.pathsep))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
24 path = os.environ.get('PATH', None)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
25 if path is None:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
26 return []
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
27 for p in os.environ.get('PATH', '').split(os.pathsep):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
28 p = os.path.join(p, str(name))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
29 if os.access(p, flags):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
30 result.append(p)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
31 for e in exts:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
32 pext = p + e
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
33 if os.access(pext, flags):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
34 result.append(pext)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
35 return result
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
36
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
37
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
38 class SSAKE:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
39 def __init__(self, logger, options):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
40 self.logger = logger
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
41 self.executables = ('SSAKE', 'makePairedOutput2EQUALfiles.pl', 'makePairedOutput2UNEQUALfiles.pl')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
42 self.logger.debug(which(self.executables[0]))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
43 self.logger.debug(which(self.executables[1]))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
44 self.logger.debug(which(self.executables[2]))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
45 self.logger.debug('Creating temp dir')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
46 self.wd = tempfile.mkdtemp()
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
47
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
48 self.kind_of_reads = int(options.kind_of_reads)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
49 if not (self.kind_of_reads):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
50 self.infile = options.if_unpaired
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
51 self.paired = 0
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
52 else:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
53 self.infile_r1 = options.if_paired_r1
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
54 self.infile_r2 = options.if_paired_r2
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
55 self.paired = 1
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
56 self.insert_size = options.insert_size
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
57 self.minnumlinks = options.minnumlinks
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
58 self.error = options.error
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
59 self.maxlinkratio = options.maxlinkratio
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
60 self.minoverlap = options.minoverlap
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
61 self.mindepthofcoverage = options.mindepthofcoverage
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
62 self.minoverlappingbases = options.minoverlappingbases
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
63 self.mincall = options.mincall
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
64 self.baseratio = options.baseratio
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
65 self.ignore_header = options.ignore_header
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
66 self.prefix = options.prefix
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
67 self.contigs = options.contigs
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
68 self.log = options.logfile
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
69 self.short = options.short
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
70 self.singlets = options.singlets
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
71 if options.seeds_file:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
72 self.seeds_file = options.seeds_file
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
73
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
74 def run(self):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
75 """ """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
76 os.chdir(self.wd)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
77 seeds = ''
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
78 if hasattr(self, 'seeds_file'):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
79 seeds = " -s %s" % self.seeds_file
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
80 if self.kind_of_reads == 1:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
81 cmd = "%s %s %s %d" % (
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
82 self.executables[1], self.infile_r1, self.infile_r2,
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
83 self.insert_size)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
84 self.logger.info("Preparing data")
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
85 execute(cmd)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
86 paired_file = "%s/paired.fa" % self.wd
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
87 command = "%s -f %s -k %d -e %s -a %s -x %d" % (self.executables[0], paired_file, self.minnumlinks, self.error, self.maxlinkratio, self.minoverlap)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
88 elif self.kind_of_reads == 2:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
89 cmd = "%s %s %s %d" % (
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
90 self.executables[2], self.infile_r1, self.infile_r2,
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
91 self.insert_size)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
92 self.logger.info("Preparing data")
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
93 execute(cmd)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
94 paired_file = "%s/paired.fa" % self.wd
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
95 unpaired_file = "%s/unpaired.fa" % self.wd
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
96 command = "%s -f %s -g %s -k %d -e %s -a %s -x %d" % (self.executables[0], paired_file, unpaired_file, self.minnumlinks, self.error, self.maxlinkratio, self.minoverlap)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
97 else:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
98 command = "%s -f %s" % (self.executables[0], self.infile)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
99 command += " %s -w %d -m %d -o %d -r %s -h %s -b %s -p %s" % (seeds, self.mindepthofcoverage, self.minoverlappingbases, self.mincall, self.baseratio, self.ignore_header, self.prefix, self.paired)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
100 self.logger.debug(command)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
101 self.logger.info("Executing SSAKE")
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
102 execute(command)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
103
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
104 with open("%s.log" % os.path.join(self.wd, self.prefix), 'rb') as ssake_log_file:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
105 self.logger.info("\n".join(["Log from SSAKE", ssake_log_file.read()]))
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
106 self.logger.info("Moving result files")
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
107 shutil.move("%s.contigs" % os.path.join(self.wd, self.prefix), self.contigs)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
108 shutil.move("%s.short" % os.path.join(self.wd, self.prefix), self.short)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
109 shutil.move("%s.singlets" % os.path.join(self.wd, self.prefix), self.singlets)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
110
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
111 def __del__(self):
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
112 shutil.rmtree(self.wd)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
113
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
114
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
115 LOG_FORMAT = '%(asctime)s|%(levelname)-8s|%(message)s'
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
116 LOG_DATEFMT = '%Y-%m-%d %H:%M:%S'
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
117 LOG_LEVELS = ['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
118
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
119
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
120 def __main__():
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
121 """ main function """
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
122 parser = optparse.OptionParser()
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
123 parser.add_option('--if_unpaired', dest='if_unpaired', help='Unpaired FASTA input file name')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
124 parser.add_option('--if_paired_r1', dest='if_paired_r1', help='Paired FASTA reads 1 input file name')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
125 parser.add_option('--if_paired_r2', dest='if_paired_r2', help='Paired FASTA reads 2 input file name')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
126 parser.add_option('-s', dest='seeds_file', help='FASTA as seeds, input file name')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
127 parser.add_option('-w', dest='mindepthofcoverage', type='int', help='minimum depth of coverage allowed for contigs')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
128 parser.add_option('-m', dest='minoverlappingbases', type='int', default=20, help='Minimum number of overlapping bases with the seed/contig during overhang consensus build up (default -m 20)')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
129 parser.add_option('-o', dest='mincall', type='int', default=2, help='mincall -o ')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
130 parser.add_option('-r', dest='baseratio', type='float', default=0.7, help='baseratio -r')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
131 parser.add_option('-k', dest='minnumlinks', type='int', default=4, help='Minimum number of links (read pairs) to compute scaffold -k')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
132 parser.add_option('-e', dest='error', type='float', default=0.75, help='Error (%) allowed on mean distance -e')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
133 parser.add_option('-a', dest='maxlinkratio', type='float', default=0.5, help='Maximum link ratio between two best contig pairs -a')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
134 parser.add_option('-x', dest='minoverlap', type='int', default=20, help='Minimum overlap required between contigs to merge adjacent contigs in a scaffold -x')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
135 parser.add_option('--ignore_header', dest='ignore_header', choices=['0', '1'], default='1', help='Ignore read name/header *will use less RAM if set to 1* -h')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
136 parser.add_option('--kind_of_reads', dest='kind_of_reads', choices=['0', '1', '2'], help='Kind of reads (-p)')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
137 parser.add_option('--iz', dest='insert_size', type='int', help='Library insert size')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
138 parser.add_option('--prefix', dest='prefix', default='ssake_pre', help='prefix')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
139 parser.add_option('--out1', dest='contigs', help='contig file')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
140 parser.add_option('--out2', dest='short', help='short file')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
141 parser.add_option('--out3', dest='singlets', help='singlets file')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
142 parser.add_option('--loglevel', choices=LOG_LEVELS, default='INFO', help='logging level (default: INFO)')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
143 parser.add_option('--logfile', help='log file (default=stderr)')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
144 (options, args) = parser.parse_args()
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
145 if len(args) > 0:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
146 parser.error('Wrong number of arguments')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
147
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
148 log_level = getattr(logging, options.loglevel)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
149 kwargs = {'format': LOG_FORMAT,
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
150 'datefmt': LOG_DATEFMT,
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
151 'level': log_level}
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
152 if options.logfile:
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
153 kwargs['filename'] = options.logfile
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
154 logging.basicConfig(**kwargs)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
155 logger = logging.getLogger('SSAKE scaffold assembly')
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
156
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
157 S = SSAKE(logger, options)
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
158 S.run()
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
159 return
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
160
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
161 if __name__ == "__main__":
0ec408bcfc80 Uploaded
crs4
parents:
diff changeset
162 __main__()