annotate gene_family_phylogeny_builder.py @ 0:f4361d941aa3 draft

Uploaded
author greg
date Tue, 11 Apr 2017 11:05:24 -0400
parents
children f73151ec679b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f4361d941aa3 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env python
f4361d941aa3 Uploaded
greg
parents:
diff changeset
2 import argparse
f4361d941aa3 Uploaded
greg
parents:
diff changeset
3 import subprocess
f4361d941aa3 Uploaded
greg
parents:
diff changeset
4
f4361d941aa3 Uploaded
greg
parents:
diff changeset
5 import utils
f4361d941aa3 Uploaded
greg
parents:
diff changeset
6
f4361d941aa3 Uploaded
greg
parents:
diff changeset
7 OUTPUT_DIR = 'phylogenomicsAnalysis_dir'
f4361d941aa3 Uploaded
greg
parents:
diff changeset
8
f4361d941aa3 Uploaded
greg
parents:
diff changeset
9 parser = argparse.ArgumentParser()
f4361d941aa3 Uploaded
greg
parents:
diff changeset
10
f4361d941aa3 Uploaded
greg
parents:
diff changeset
11 parser.add_argument('--alignments_type', dest='alignments_type', help='Input alignments type produced by the GeneFamilyAligner')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
12 parser.add_argument('--bootstrap_replicates', dest='bootstrap_replicates', type=int, default=None, help='Number of replicates for rapid bootstrap analysis')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
13 parser.add_argument('--config_dir', dest='config_dir', help='Directory containing default configuration files')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
14 parser.add_argument('--max_orthogroup_size', dest='max_orthogroup_size', type=int, help='Maximum number of sequences in orthogroup alignments')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
15 parser.add_argument('--method', dest='method', help='Protein clustering method')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
16 parser.add_argument('--min_orthogroup_size', dest='min_orthogroup_size', type=int, help='Minimum number of sequences in orthogroup alignments')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
17 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of threads to use for execution')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
18 parser.add_argument('--orthogroup_aln', dest='orthogroup_aln', help="Input dataset files_path")
f4361d941aa3 Uploaded
greg
parents:
diff changeset
19 parser.add_argument('--output', dest='output', help='Output for phylogenetic trees')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
20 parser.add_argument('--output_dir', dest='output_dir', help='output.files_path')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
21 parser.add_argument('--rooting_order', dest='rooting_order', default=None, help='Rooting order configuration for rooting trees')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
22 parser.add_argument('--scaffold', dest='scaffold', help='Orthogroups or gene families proteins scaffold')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
23 parser.add_argument('--sequence_type', dest='sequence_type', help="Sequence type used in the phylogenetic inference")
f4361d941aa3 Uploaded
greg
parents:
diff changeset
24 parser.add_argument('--tree_inference', dest='tree_inference', help='Phylogenetic trees inference method')
f4361d941aa3 Uploaded
greg
parents:
diff changeset
25
f4361d941aa3 Uploaded
greg
parents:
diff changeset
26 args = parser.parse_args()
f4361d941aa3 Uploaded
greg
parents:
diff changeset
27
f4361d941aa3 Uploaded
greg
parents:
diff changeset
28 # Build the command line.
f4361d941aa3 Uploaded
greg
parents:
diff changeset
29 cmd = 'GeneFamilyPhylogenyBuilder'
f4361d941aa3 Uploaded
greg
parents:
diff changeset
30 cmd += ' --alignment_type %s' % args.tree_inference
f4361d941aa3 Uploaded
greg
parents:
diff changeset
31 if args.bootstrap_replicates is not None:
f4361d941aa3 Uploaded
greg
parents:
diff changeset
32 cmd += ' --bootstrap_replicates %d' % args.bootstrap_replicates
f4361d941aa3 Uploaded
greg
parents:
diff changeset
33 cmd += ' --config_dir %s' % args.config_dir
f4361d941aa3 Uploaded
greg
parents:
diff changeset
34 cmd += ' --max_orthogroup_size %d' % args.max_orthogroup_size
f4361d941aa3 Uploaded
greg
parents:
diff changeset
35 cmd += ' --method %s' % args.method
f4361d941aa3 Uploaded
greg
parents:
diff changeset
36 cmd += ' --min_orthogroup_size %d' % args.min_orthogroup_size
f4361d941aa3 Uploaded
greg
parents:
diff changeset
37 cmd += ' --num_threads %d' % args.num_threads
f4361d941aa3 Uploaded
greg
parents:
diff changeset
38 cmd += ' --orthogroup_aln %s' % args.orthogroup_aln
f4361d941aa3 Uploaded
greg
parents:
diff changeset
39 if args.rooting_order is not None:
f4361d941aa3 Uploaded
greg
parents:
diff changeset
40 cmd += ' --rooting_order %s' % args.rooting_order
f4361d941aa3 Uploaded
greg
parents:
diff changeset
41 cmd += ' --scaffold %s' % args.scaffold
f4361d941aa3 Uploaded
greg
parents:
diff changeset
42 cmd += ' --sequence_type %s' % args.sequence_type
f4361d941aa3 Uploaded
greg
parents:
diff changeset
43 cmd += ' --tree_inference %s' % args.tree_inference
f4361d941aa3 Uploaded
greg
parents:
diff changeset
44 # Run the command.
f4361d941aa3 Uploaded
greg
parents:
diff changeset
45 proc = subprocess.Popen(args=cmd, stderr=subprocess.PIPE, stdout=subprocess.PIPE, shell=True)
f4361d941aa3 Uploaded
greg
parents:
diff changeset
46 rc = proc.wait()
f4361d941aa3 Uploaded
greg
parents:
diff changeset
47 utils.check_execution_errors(rc, proc.stderr)
f4361d941aa3 Uploaded
greg
parents:
diff changeset
48 utils.move_directory_files(OUTPUT_DIR, args.output_dir)
f4361d941aa3 Uploaded
greg
parents:
diff changeset
49 utils.write_html_output(args.output, 'Phylogenetic trees', args.output_dir)