Mercurial > repos > greg > gene_family_aligner
comparison gene_family_aligner.py @ 0:6accbd3a1449 draft
Uploaded
| author | greg |
|---|---|
| date | Fri, 07 Apr 2017 13:00:29 -0400 |
| parents | |
| children | 01c08c35d051 |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:6accbd3a1449 |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 import argparse | |
| 3 import subprocess | |
| 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 trimAl's 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=None, type=float, help='Remove sites in alignments with gaps of') | |
| 14 parser.add_argument('--iterative_realignment', dest='iterative_realignment', type=int, default=None, help='"Maximum number of iterations') | |
| 15 parser.add_argument('--method', dest='method', help='Protein clustering method') | |
| 16 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of threads to use for execution') | |
| 17 parser.add_argument('--orthogroup_faa', dest='orthogroup_faa', help="Directory of input fasta datasets") | |
| 18 parser.add_argument('--output', dest='output', help="Output dataset") | |
| 19 parser.add_argument('--output_dir', dest='output_dir', help="Output dataset file_path directory") | |
| 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=None, type=float, help='Remove sequences with gaps of') | |
| 23 parser.add_argument('--scaffold', dest='scaffold', help='Orthogroups or gene families proteins scaffold') | |
| 24 | |
| 25 args = parser.parse_args() | |
| 26 | |
| 27 # Build the command line. | |
| 28 cmd = 'GeneFamilyAligner' | |
| 29 cmd += ' --orthogroup_faa %s' % args.orthogroup_faa | |
| 30 cmd += ' --scaffold %s' % args.scaffold | |
| 31 cmd += ' --method %s' % args.method | |
| 32 cmd += ' --alignment_method %s' % args.alignment_method | |
| 33 if args.alignment_method == 'pasta': | |
| 34 if args.pasta_script_path is not None: | |
| 35 cmd += ' --pasta_script_path %s' % args.pasta_script_path | |
| 36 if args.pasta_iter_limit is not None: | |
| 37 cmd += ' --pasta_iter_limit %d' % args.pasta_iter_limit | |
| 38 cmd += ' --num_threads %d' % args.num_threads | |
| 39 if args.orthogroup_fna is not None: | |
| 40 cmd += ' --orthogroup_fna' | |
| 41 if args.automated_trimming is not None: | |
| 42 cmd += ' --automated_trimming' | |
| 43 if args.gap_trimming is not None: | |
| 44 cmd += ' --gap_trimming %4f' % args.gap_trimming | |
| 45 if args.remove_sequences is not None: | |
| 46 cmd += ' --remove_sequences %4f' % args.remove_sequences | |
| 47 if args.iterative_realignment is not None: | |
| 48 cmd += ' --iterative_realignment %d' % args.iterative_realignment | |
| 49 # Run the command. | |
| 50 proc = subprocess.Popen(args=cmd, stderr=subprocess.PIPE, stdout=subprocess.PIPE, shell=True) | |
| 51 rc = proc.wait() | |
| 52 utils.check_execution_errors(rc, proc.stderr) | |
| 53 utils.move_directory_files(OUTPUT_DIR, args.output_dir) | |
| 54 utils.write_html_output(args.output, 'Aligned gene family sequences', args.output_dir) |