Mercurial > repos > peterjc > mira4_assembler

comparison tools/mira4_0/mira4_make_bam.py @ 2:4eb32a3d67d1 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
v0.0.8 - renamed folder, added note about mirabait
author peterjc
date Wed, 02 Sep 2015 07:46:29 -0400
parents
children 1713289d9908
comparison
equal deleted inserted replaced
1:70248e6e3efc 2:4eb32a3d67d1
1 #!/usr/bin/env python
2 """Wrapper script using miraconvert & samtools to get BAM from MIRA.
3 """
4 import os
5 import sys
6 import shutil
7 import subprocess
8 import tempfile
9
10 def sys_exit(msg, err=1):
11 sys.stderr.write(msg+"\n")
12 sys.exit(err)
13
14 def run(cmd, log_handle):
15 try:
16 child = subprocess.Popen(cmd, shell=True,
17 stdout=subprocess.PIPE,
18 stderr=subprocess.STDOUT)
19 except Exception, err:
20 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
21 #TODO - call clean up?
22 log_handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
23 sys.exit(1)
24 #Use .communicate as can get deadlocks with .wait(),
25 stdout, stderr = child.communicate()
26 assert not stderr #Should be empty as sent to stdout
27 if len(stdout) > 10000:
28 #miraconvert can be very verbose (is holding stdout in RAM a problem?)
29 stdout = stdout.split("\n")
30 stdout = stdout[:10] + ["...", "<snip>", "..."] + stdout[-10:]
31 stdout = "\n".join(stdout)
32 log_handle.write(stdout)
33 return child.returncode
34
35 def depad(fasta_file, sam_file, bam_file, log_handle):
36 log_handle.write("\n================= Converting MIRA assembly from SAM to BAM ===================\n")
37 #Also doing SAM to (uncompressed) BAM during depad
38 bam_stem = bam_file + ".tmp" # Have write permissions and want final file in this folder
39 cmd = 'samtools depad -S -u -T "%s" "%s" | samtools sort - "%s"' % (fasta_file, sam_file, bam_stem)
40 return_code = run(cmd, log_handle)
41 if return_code:
42 return "Error %i from command:\n%s" % (return_code, cmd)
43 if not os.path.isfile(bam_stem + ".bam"):
44 return "samtools depad or sort failed to produce BAM file"
45
46 log_handle.write("\n====================== Indexing MIRA assembly BAM file =======================\n")
47 cmd = 'samtools index "%s.bam"' % bam_stem
48 return_code = run(cmd, log_handle)
49 if return_code:
50 return "Error %i from command:\n%s" % (return_code, cmd)
51 if not os.path.isfile(bam_stem + ".bam.bai"):
52 return "samtools indexing of BAM file failed to produce BAI file"
53
54 shutil.move(bam_stem + ".bam", bam_file)
55 os.remove(bam_stem + ".bam.bai") #Let Galaxy handle that...
56
57
58 def make_bam(mira_convert, maf_file, fasta_file, bam_file, log_handle):
59 if not os.path.isfile(mira_convert):
60 return "Missing binary %r" % mira_convert
61 if not os.path.isfile(maf_file):
62 return "Missing input MIRA file: %r" % maf_file
63 if not os.path.isfile(fasta_file):
64 return "Missing padded FASTA file: %r" % fasta_file
65
66 log_handle.write("\n====================== Converting MIRA assembly to SAM =======================\n")
67 tmp_dir = tempfile.mkdtemp()
68 sam_file = os.path.join(tmp_dir, "x.sam")
69
70 # Note add nbb to the template name, possible MIRA 4.0 RC4 bug
71 cmd = '"%s" -f maf -t samnbb "%s" "%snbb"' % (mira_convert, maf_file, sam_file)
72 return_code = run(cmd, log_handle)
73 if return_code:
74 return "Error %i from command:\n%s" % (return_code, cmd)
75 if not os.path.isfile(sam_file):
76 return "Conversion from MIRA to SAM failed"
77
78 #Also doing SAM to (uncompressed) BAM during depad
79 msg = depad(fasta_file, sam_file, bam_file, log_handle)
80 if msg:
81 return msg
82
83 os.remove(sam_file)
84 os.rmdir(tmp_dir)
85
86 return None #Good :)
87
88 if __name__ == "__main__":
89 mira_convert, maf_file, fasta_file, bam_file = sys.argv[1:]
90 msg = make_bam(mira_convert, maf_file, fasta_file, bam_file, sys.stdout)
91 if msg:
92 sys_exit(msg)