|
0
|
1 #!/usr/bin/env python
|
|
|
2 import argparse
|
|
|
3 import os
|
|
|
4
|
|
|
5 import utils
|
|
|
6
|
|
|
7 OUTPUT_DIR = 'geneFamilyAlignments_dir'
|
|
|
8
|
|
|
9 parser = argparse.ArgumentParser()
|
|
|
10 parser.add_argument('--alignment_method', dest='alignment_method', help='Multiple sequence alignments method')
|
|
|
11 parser.add_argument('--automated_trimming', dest='automated_trimming', default=None, help='Trims alignments using trimAls ML heuristic trimming approach')
|
|
|
12 parser.add_argument('--codon_alignments', dest='codon_alignments', default=None, help='Flag for constructing orthogroup multiple codon alignments')
|
|
|
13 parser.add_argument('--gap_trimming', dest='gap_trimming', default=0, type=float, help='Remove sites in alignments with gaps of')
|
|
|
14 parser.add_argument('--iterative_realignment', dest='iterative_realignment', type=int, default=0, help='Maximum number of iterations')
|
|
|
15 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of threads to use for execution')
|
|
|
16 parser.add_argument('--orthogroup_faa', dest='orthogroup_faa', help='Directory of input fasta datasets')
|
|
|
17 parser.add_argument('--output', dest='output', help='Output dataset')
|
|
|
18 parser.add_argument('--output_dir', dest='output_dir', help='Output dataset files_path directory')
|
|
1
|
19 parser.add_argument('--output_dataset_collection', dest='output_dataset_collection', default=None, help='Output additional dataset collection of files.')
|
|
0
|
20 parser.add_argument('--pasta_iter_limit', dest='pasta_iter_limit', type=int, default=None, help='Maximum number of iteration that the PASTA algorithm will execute')
|
|
|
21 parser.add_argument('--pasta_script_path', dest='pasta_script_path', default=None, help='Path to script for executing pasta')
|
|
|
22 parser.add_argument('--remove_sequences', dest='remove_sequences', default=0, type=float, help='Remove sequences with gaps of')
|
|
|
23
|
|
|
24 args = parser.parse_args()
|
|
|
25
|
|
|
26 # Build the command line.
|
|
|
27 cmd = 'GeneFamilyAligner'
|
|
|
28 cmd += ' --orthogroup_faa %s' % args.orthogroup_faa
|
|
|
29 cmd += ' --alignment_method %s' % args.alignment_method
|
|
|
30 if args.alignment_method == 'pasta':
|
|
|
31 if args.pasta_script_path is not None:
|
|
|
32 cmd += ' --pasta_script_path %s' % args.pasta_script_path
|
|
|
33 if args.pasta_iter_limit is not None:
|
|
|
34 cmd += ' --pasta_iter_limit %d' % args.pasta_iter_limit
|
|
|
35 cmd += ' --num_threads %d' % args.num_threads
|
|
|
36 if args.codon_alignments is not None:
|
|
|
37 cmd += ' --codon_alignments'
|
|
|
38 if args.automated_trimming is not None:
|
|
|
39 cmd += ' --automated_trimming'
|
|
|
40 if args.gap_trimming > 0:
|
|
|
41 cmd += ' --gap_trimming %4f' % args.gap_trimming
|
|
|
42 if args.remove_sequences > 0:
|
|
|
43 cmd += ' --remove_sequences %4f' % args.remove_sequences
|
|
|
44 if args.iterative_realignment > 0:
|
|
|
45 cmd += ' --iterative_realignment %d' % args.iterative_realignment
|
|
|
46
|
|
|
47 # Run the command.
|
|
|
48 utils.run_command(cmd)
|
|
|
49
|
|
|
50 # Handle outputs.
|
|
|
51 if args.codon_alignments is None:
|
|
|
52 src_output_dir = OUTPUT_DIR
|
|
|
53 else:
|
|
|
54 src_output_dir = os.path.join(OUTPUT_DIR, 'orthogroups_aln')
|
|
1
|
55 if args.output_dataset_collection is not None:
|
|
|
56 utils.move_directory_files(src_output_dir, args.output_dataset_collection, copy=True)
|
|
0
|
57 utils.move_directory_files(src_output_dir, args.output_dir)
|
|
|
58 utils.write_html_output(args.output, 'Aligned gene family sequences', args.output_dir)
|
