Mercurial > repos > rmarenco > multi_fasta_glimmer_hmm

annotate multi_glimmer.py @ 0:0ddb5ee32ff6 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
author rmarenco
date Thu, 18 Aug 2016 18:50:00 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
1 #!/usr/bin/python
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
2 # -*- coding: utf8 -*-
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
3
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
4 import argparse
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
5 import os
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
6 import subprocess
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
7 import sys
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
8
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
9
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
10 def main():
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
11 parser = argparse.ArgumentParser(description='Get a multi-fasta, the trained_dir and the output file as inputs, '
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
12 'to generate GlimmerHMM gene prediction over all contigs')
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
13
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
14 parser.add_argument('--multi_fasta', help='Multi fasta file to run GlimmerHMM on', required=True)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
15
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
16 parser.add_argument('--trained_dir', help='Path to the GlimmerHMM trained_dir', required=True)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
17
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
18 parser.add_argument('--output', help='file to output the result into', required=True)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
19
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
20 args = parser.parse_args()
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
21
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
22 multi_fasta = args.multi_fasta
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
23 trained_dir = args.trained_dir
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
24 # TODO: Temporary fix for the issue with config.file in human/. Next: GC Content to select the appropriate folder
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
25 if trained_dir.split('/')[-1] == "human":
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
26 trained_dir = os.path.join(trained_dir, "Train0-43")
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
27
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
28 output_file = args.output
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
29 temp_contig = "temp_contig"
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
30
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
31 def exec_glimmer(contig_file, first_time=False):
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
32 p = subprocess.Popen(["glimmerhmm", contig_file, trained_dir, "-g"],
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
33 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
34 output, errors = p.communicate()
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
35
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
36 p.wait()
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
37 # Process the error if != "Done"
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
38 if not errors or (errors.split()[0] != "Done"):
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
39 raise Exception("Error in glimmer: {0}".format(errors))
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
40 else:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
41 sys.stdout.write(errors)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
42 # If not first time, we need to remove the first comments
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
43 if not first_time:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
44 output = "\n".join(output.split("\n")[1:])
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
45
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
46 return output
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
47
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
48 with open(output_file, 'w+') as o:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
49 with open(multi_fasta, 'r') as mf:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
50 is_first_time = True
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
51 for i, line in enumerate(mf):
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
52 if line[0] == '>':
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
53 # If it is the first time we finish to read a contig, we let glimmer add the full comments
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
54 # and write into the output the result
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
55 if is_first_time is True and i != 0:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
56 o.write(exec_glimmer(temp_contig, first_time=is_first_time))
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
57 is_first_time = False
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
58 # Else we call glimmer and say this is not the first time (so remove the first comment)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
59 # and dump into the output file the result
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
60 elif i > 0:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
61 o.write(exec_glimmer(temp_contig))
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
62
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
63 # Because we are on an indication of a beginning of a sequence, we need to create an empty file
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
64 # to dump the line into
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
65 with open(temp_contig, 'w+') as tc:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
66 tc.write(line)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
67 else:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
68 # We are in the sequence of a contig, so we append the line in the file
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
69 with open(temp_contig, 'a+') as tc:
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
70 tc.write(line)
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
71 # The file is terminate, we did read another contig so we need to save this last one
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
72 o.write(exec_glimmer(temp_contig, first_time=is_first_time))
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
73
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
74 if __name__ == "__main__":
0ddb5ee32ff6 planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 28bd73b26b50165eded1d9ba995979acdf005ad1-dirty
rmarenco
parents:
diff changeset
75 main()