Mercurial > repos > greg > gene_family_classifier

comparison gene_family_classifier.py @ 141:07bf26b7a1ab draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Fri, 05 May 2017 09:32:21 -0400
parents b2ae23e484e8
children 36ed0037292a
comparison
equal deleted inserted replaced
140:000b2401eae2 141:07bf26b7a1ab
1 #!/usr/bin/env python 1 #!/usr/bin/env python
2 import argparse 2 import argparse
3 import os 3 import os
4 import shutil 4 import shutil
5 import subprocess
6 5
7 import utils 6 import utils
8 7
9 8
10 BUFF_SIZE = 1048576 9 BUFF_SIZE = 1048576
11 OUTPUT_DIR = 'geneFamilyClassification_dir' 10 OUTPUT_DIR = 'geneFamilyClassification_dir'
12 11
13 parser = argparse.ArgumentParser() 12 parser = argparse.ArgumentParser()
14 parser.add_argument('--input', dest='input', help="Input dataset") 13 parser.add_argument('--input', dest='input', help='Input dataset')
15 parser.add_argument('--scaffold', dest='scaffold', help='Orthogroups or gene families proteins scaffold') 14 parser.add_argument('--scaffold', dest='scaffold', help='Orthogroups or gene families proteins scaffold')
16 parser.add_argument('--method', dest='method', help='Protein clustering method') 15 parser.add_argument('--method', dest='method', help='Protein clustering method')
17 parser.add_argument('--classifier', dest='classifier', help='Protein classification method') 16 parser.add_argument('--classifier', dest='classifier', help='Protein classification method')
18 parser.add_argument('--config_dir', dest='config_dir', help='Directory containing default configuration files') 17 parser.add_argument('--config_dir', dest='config_dir', help='Directory containing default configuration files')
19 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of threads to use for execution') 18 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of threads to use for execution')
20 parser.add_argument('--super_orthogroups', dest='super_orthogroups', default=None, help='Super orthogroups clustering specification') 19 parser.add_argument('--super_orthogroups', dest='super_orthogroups', default=None, help='Super orthogroups clustering specification')
21 parser.add_argument('--single_copy_custom', dest='single_copy_custom', default=None, help='Custom single copy orthogroup configuration') 20 parser.add_argument('--single_copy_custom', dest='single_copy_custom', default=None, help='Custom single copy orthogroup configuration')
22 parser.add_argument('--single_copy_taxa', dest='single_copy_taxa', type=int, default=0, help='"Minimum single copy taxa required in orthogroup') 21 parser.add_argument('--single_copy_taxa', dest='single_copy_taxa', type=int, default=0, help='Minimum single copy taxa required in orthogroup')
23 parser.add_argument('--taxa_present', dest='taxa_present', type=int, default=0, help='Minimum taxa required in single copy orthogroup') 22 parser.add_argument('--taxa_present', dest='taxa_present', type=int, default=0, help='Minimum taxa required in single copy orthogroup')
24 parser.add_argument('--orthogroup_fasta', dest='orthogroup_fasta', default=None, help='Flag to create orthogroup sequences') 23 parser.add_argument('--orthogroup_fasta', dest='orthogroup_fasta', default=None, help='Flag to create orthogroup sequences')
25 parser.add_argument('--coding_sequences', dest='coding_sequences', default=None, help='Flag to create orthogroup coding sequences') 24 parser.add_argument('--coding_sequences', dest='coding_sequences', default=None, help='Flag to create orthogroup coding sequences')
26 parser.add_argument('--save_hmmscan_log', dest='save_hmmscan_log', default=None, help='Flag to save the hmmscan log') 25 parser.add_argument('--save_hmmscan_log', dest='save_hmmscan_log', default=None, help='Flag to save the hmmscan log')
27 parser.add_argument('--hmmscan_log', dest='hmmscan_log', default=None, help='hmmscan log file') 26 parser.add_argument('--hmmscan_log', dest='hmmscan_log', default=None, help='hmmscan log file')
31 parser.add_argument('--output_ptorthocs_dir', dest='output_ptorthocs_dir', default=None, help='output_ptorthocs.files_path') 30 parser.add_argument('--output_ptorthocs_dir', dest='output_ptorthocs_dir', default=None, help='output_ptorthocs.files_path')
32 parser.add_argument('--output_ptsco', dest='output_ptsco', default=None, help='Output for single copy orthogroups') 31 parser.add_argument('--output_ptsco', dest='output_ptsco', default=None, help='Output for single copy orthogroups')
33 parser.add_argument('--output_ptsco_dir', dest='output_ptsco_dir', default=None, help='output_ptsco.files_path') 32 parser.add_argument('--output_ptsco_dir', dest='output_ptsco_dir', default=None, help='output_ptsco.files_path')
34 33
35 args = parser.parse_args() 34 args = parser.parse_args()
36
37 35
38 # Build the command line. 36 # Build the command line.
39 cmd = 'GeneFamilyClassifier' 37 cmd = 'GeneFamilyClassifier'
40 cmd += ' --proteins %s' % args.input 38 cmd += ' --proteins %s' % args.input
41 cmd += ' --scaffold %s' % args.scaffold 39 cmd += ' --scaffold %s' % args.scaffold
59 if args.coding_sequences is None: 57 if args.coding_sequences is None:
60 create_corresponding_coding_sequences = False 58 create_corresponding_coding_sequences = False
61 else: 59 else:
62 create_corresponding_coding_sequences = True 60 create_corresponding_coding_sequences = True
63 cmd += ' --coding_sequences %s' % args.coding_sequences 61 cmd += ' --coding_sequences %s' % args.coding_sequences
62
64 # Run the command. 63 # Run the command.
65 proc = subprocess.Popen(args=cmd, stderr=subprocess.PIPE, stdout=subprocess.PIPE, shell=True) 64 utils.run_command(cmd)
66 rc = proc.wait() 65
67 utils.check_execution_errors(rc, proc.stderr)
68 # Handle hmmscan.log output. 66 # Handle hmmscan.log output.
69 if args.classifier in ['hmmscan', 'both']: 67 if args.classifier in ['hmmscan', 'both']:
70 src_hmmscan_log = os.path.join(OUTPUT_DIR, 'hmmscan.log') 68 src_hmmscan_log = os.path.join(OUTPUT_DIR, 'hmmscan.log')
71 if os.path.exists(src_hmmscan_log): 69 if os.path.exists(src_hmmscan_log):
72 if args.save_hmmscan_log is None: 70 if args.save_hmmscan_log is None:
73 os.remove(src_hmmscan_log) 71 os.remove(src_hmmscan_log)
74 else: 72 else:
75 shutil.move(src_hmmscan_log, args.hmmscan_log) 73 shutil.move(src_hmmscan_log, args.hmmscan_log)
74
76 # Handle orthogroups outputs. 75 # Handle orthogroups outputs.
77 if create_ortho_sequences: 76 if create_ortho_sequences:
78 if create_corresponding_coding_sequences: 77 if create_corresponding_coding_sequences:
79 out_file = args.output_ptorthocs 78 out_file = args.output_ptorthocs
80 orthogroups_fasta_dest_dir = args.output_ptorthocs_dir 79 orthogroups_fasta_dest_dir = args.output_ptorthocs_dir
84 orthogroups_fasta_dest_dir = args.output_ptortho_dir 83 orthogroups_fasta_dest_dir = args.output_ptortho_dir
85 title = 'Orthogroups files' 84 title = 'Orthogroups files'
86 orthogroups_fasta_src_dir = os.path.join(OUTPUT_DIR, 'orthogroups_fasta') 85 orthogroups_fasta_src_dir = os.path.join(OUTPUT_DIR, 'orthogroups_fasta')
87 utils.move_directory_files(orthogroups_fasta_src_dir, orthogroups_fasta_dest_dir) 86 utils.move_directory_files(orthogroups_fasta_src_dir, orthogroups_fasta_dest_dir)
88 utils.write_html_output(out_file, title, orthogroups_fasta_dest_dir) 87 utils.write_html_output(out_file, title, orthogroups_fasta_dest_dir)
88
89 # Handle single copy orthogroup outputs. 89 # Handle single copy orthogroup outputs.
90 if args.output_ptsco is not None: 90 if args.output_ptsco is not None:
91 single_copy_fasta_src_dir = os.path.join(OUTPUT_DIR, 'single_copy_fasta') 91 single_copy_fasta_src_dir = os.path.join(OUTPUT_DIR, 'single_copy_fasta')
92 single_copy_fasta_dest_dir = args.output_ptsco_dir 92 single_copy_fasta_dest_dir = args.output_ptsco_dir
93 title = 'Single copy orthogroups files' 93 title = 'Single copy orthogroups files'