Mercurial > repos > peterjc > mira_assembler

annotate tools/sr_assembly/mira.py @ 1:e53a79816f5f

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
v0.0.2 - Improve capture of stdout/stderr (should see it as it runs)
author peterjc
date Thu, 16 Jun 2011 04:44:00 -0400
parents 03b240624b5a
children 298f5c1d9521
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2 """A simple wrapper script to call MIRA and collect its output.
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3 """
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 import os
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5 import sys
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 import subprocess
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 import shutil
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8 import time
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
9
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
10 def stop_err(msg, err=1):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
11 sys.stderr.write(msg+"\n")
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
12 sys.exit(err)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14 def tcs_to_tabular(old, new):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
15 in_handle = open(old, "rU")
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
16 out_handle = open(new, "w")
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
17 assert in_handle.readline() == "#TCS V1.0\n"
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
18 assert in_handle.readline() == "#\n"
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
19 assert in_handle.readline() == "# contig name padPos upadPos | B Q | tcov covA covC covG covT cov* | qA qC qG qT q* | S | Tags\n"
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
20 assert in_handle.readline() == "#\n"
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
21 out_handle.write("#%s\n" % "\t".join(["contig", "pasPos", "upadPos", "B", "Q",
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
22 "tcov", "covA", "covC", "covG", "covT", "cov*",
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
23 "qA", "qC", "qG", "qT", "q*", "S", "Tags"]))
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
24 for line in in_handle:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
25 parts = line.rstrip("\n").split(None,22)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
26 assert parts[3] == parts[6] == parts[13] == parts[19] == parts[21] == "|"
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
27 wanted = parts[:3] + parts[4:6]+parts[7:13]+parts[14:19]+parts[20:21]+parts[22:]
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
28 out_handle.write("%s\n" % "\t".join(wanted))
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
29 out_handle.close()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
30 in_handle.close()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
31
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
32 def collect_output(temp, name):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
33 n3 = (temp, name, name, name)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
34 tcs_to_tabular("%s/%s_assembly/%s_d_results/%s_out.tcs" % n3, out_tcs)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
35 for old, new in [("%s/%s_assembly/%s_d_results/%s_out.unpadded.fasta" % n3, out_fasta),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
36 ("%s/%s_assembly/%s_d_results/%s_out.unpadded.fasta.qual" % n3, out_qual),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
37 ("%s/%s_assembly/%s_d_results/%s_out.wig" % n3, out_wig),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
38 ("%s/%s_assembly/%s_d_results/%s_out.caf" % n3, out_caf),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
39 ("%s/%s_assembly/%s_d_results/%s_out.ace" % n3, out_ace)]:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
40 if not os.path.isfile(old):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
41 stop_err("Missing %s output file" % os.path.splitext(old)[-1])
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
42 else:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
43 shutil.move(old, new)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
44
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
45 def clean_up(temp, name):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
46 folder = "%s/%s_assembly" % (temp, name)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
47 if os.path.isdir(folder):
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
48 shutil.rmtree(folder)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
49
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
50 #TODO - Run MIRA in /tmp or a configurable directory?
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
51 temp = "."
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
52 name, out_fasta, out_qual, out_tcs, out_ace, out_caf, out_wig, out_log = sys.argv[1:9]
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
53 start_time = time.time()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
54 try:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
55 cmd = " ".join(sys.argv[9:])
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
56 child = subprocess.Popen(sys.argv[9:],
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
57 stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
58 except Exception, err:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
59 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
60 sys.exit(1)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
61 #Use .communicate as can get deadlocks with .wait(),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
62 stdout, stderr = child.communicate()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
63 run_time = time.time() - start_time
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
64 return_code = child.returncode
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
65 handle = open(out_log, "w")
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
66 handle.write(stdout)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
67 handle.write("\n\nMIRA took %0.2f minutes\n" % (run_time / 60.0))
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
68 print "MIRA took %0.2f minutes" % (run_time / 60.0)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
69 if return_code:
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
70 handle.write("Return error code %i from command:\n" % return_code)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
71 handle.write(cmd + "\n")
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 handle.close()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 clean_up(temp, name)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 stop_err("Return error code %i from command:\n%s" % (return_code, cmd),
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 return_code)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 handle.close()
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 collect_output(temp, name)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 clean_up(temp, name)
03b240624b5a Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
80 print "Done"