annotate tools/mira4_0/mira4_make_bam.py @ 35:259891fce7fd draft

planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
author peterjc
date Wed, 09 Aug 2017 13:32:11 -0400
parents 0785a6537f3e
children cee8f9005e43
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
2 """Wrapper script using miraconvert & samtools to get BAM from MIRA.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
3 """
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
4
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
5 import os
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
6 import shutil
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
7 import subprocess
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
8 import sys
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
9 import tempfile
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
10
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
11
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
12 def run(cmd, log_handle):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
13 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
14 child = subprocess.Popen(cmd, shell=True,
34
0785a6537f3e planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 6405ba93fcec7ea93452bf54d559c7507ee7a57c
peterjc
parents: 32
diff changeset
15 universal_newlines=True,
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
16 stdout=subprocess.PIPE,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
17 stderr=subprocess.STDOUT)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
18 except Exception as err:
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
19 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
20 # TODO - call clean up?
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
21 log_handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
22 sys.exit(1)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
23 # Use .communicate as can get deadlocks with .wait(),
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
24 stdout, stderr = child.communicate()
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
25 assert not stderr # Should be empty as sent to stdout
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
26 if len(stdout) > 10000:
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
27 # miraconvert can be very verbose (is holding stdout in RAM a problem?)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
28 stdout = stdout.split("\n")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
29 stdout = stdout[:10] + ["...", "<snip>", "..."] + stdout[-10:]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
30 stdout = "\n".join(stdout)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
31 log_handle.write(stdout)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
32 return child.returncode
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
33
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
34
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
35 def depad(fasta_file, sam_file, bam_file, log_handle):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
36 log_handle.write("\n================= Converting MIRA assembly from SAM to BAM ===================\n")
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
37 # Also doing SAM to (uncompressed) BAM during depad
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
38 bam_stem = bam_file + ".tmp" # Have write permissions and want final file in this folder
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
39 cmd = 'samtools depad -S -u -T "%s" "%s" | samtools sort - "%s"' % (fasta_file, sam_file, bam_stem)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
40 return_code = run(cmd, log_handle)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
41 if return_code:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
42 return "Error %i from command:\n%s" % (return_code, cmd)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
43 if not os.path.isfile(bam_stem + ".bam"):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
44 return "samtools depad or sort failed to produce BAM file"
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
45
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
46 log_handle.write("\n====================== Indexing MIRA assembly BAM file =======================\n")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
47 cmd = 'samtools index "%s.bam"' % bam_stem
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
48 return_code = run(cmd, log_handle)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
49 if return_code:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
50 return "Error %i from command:\n%s" % (return_code, cmd)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
51 if not os.path.isfile(bam_stem + ".bam.bai"):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
52 return "samtools indexing of BAM file failed to produce BAI file"
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
53
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
54 shutil.move(bam_stem + ".bam", bam_file)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
55 os.remove(bam_stem + ".bam.bai") # Let Galaxy handle that...
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
56
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
57
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
58 def make_bam(mira_convert, maf_file, fasta_file, bam_file, log_handle):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
59 if not os.path.isfile(maf_file):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
60 return "Missing input MIRA file: %r" % maf_file
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
61 if not os.path.isfile(fasta_file):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
62 return "Missing padded FASTA file: %r" % fasta_file
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
63
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
64 log_handle.write("\n====================== Converting MIRA assembly to SAM =======================\n")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
65 tmp_dir = tempfile.mkdtemp()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
66 sam_file = os.path.join(tmp_dir, "x.sam")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
67
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
68 # Note add nbb to the template name, possible MIRA 4.0 RC4 bug
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
69 cmd = '"%s" -f maf -t samnbb "%s" "%snbb"' % (mira_convert, maf_file, sam_file)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
70 return_code = run(cmd, log_handle)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
71 if return_code:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
72 return "Error %i from command:\n%s" % (return_code, cmd)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
73 if not os.path.isfile(sam_file):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
74 return "Conversion from MIRA to SAM failed"
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
75
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
76 # Also doing SAM to (uncompressed) BAM during depad
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
77 msg = depad(fasta_file, sam_file, bam_file, log_handle)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
78 if msg:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
79 return msg
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
80
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
81 os.remove(sam_file)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
82 os.rmdir(tmp_dir)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
83
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
84 return None # Good :)
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
85
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
86
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
87 if __name__ == "__main__":
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
88 mira_convert, maf_file, fasta_file, bam_file = sys.argv[1:]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
89 msg = make_bam(mira_convert, maf_file, fasta_file, bam_file, sys.stdout)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
90 if msg:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
91 sys.exit(msg)