annotate assembly_post_processor.py @ 58:e7f264e21d9a draft

Uploaded
author greg
date Mon, 22 May 2017 10:37:52 -0400
parents ee5edbfd99c3
children baf10f7f3d4a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env python
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
2 import argparse
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
3 import os
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
4
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
5 import utils
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
6
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
7 OUTPUT_DIR = 'assemblyPostProcessing_dir'
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
8
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
9 parser = argparse.ArgumentParser()
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
10 parser.add_argument('--dereplicate', dest='dereplicate', default=None, help='Remove duplicate sequences')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
11 parser.add_argument('--gap_trimming', dest='gap_trimming', type=float, default=0, help='Trim alignments')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
12 parser.add_argument('--gene_family_search', dest='gene_family_search', default=None, help='Targeted gene families')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
13 parser.add_argument('--method', dest='method', default=None, help='Protein clustering method')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
14 parser.add_argument('--min_length', dest='min_length', type=int, default=0, help='Minimum sequence length')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
15 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of processors')
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
16 parser.add_argument('--output_pttgf', dest='output_pttgf', default=None, help='Primary targeted gene families dataset')
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
17 parser.add_argument('--output_pttgf_dir', dest='output_pttgf_dir', default=None, help='Directory hierarchy of targeted gene family datasets')
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
18 parser.add_argument('--prediction_method', dest='prediction_method', help='Coding regions prediction method')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
19 parser.add_argument('--scaffold', dest='scaffold', default=None, help='Gene family scaffold')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
20 parser.add_argument('--score_matrices', dest='score_matrices', default=None, help='Scores matrices')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
21 parser.add_argument('--strand_specific', dest='strand_specific', default=None, help='Strand-specific assembly')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
22 parser.add_argument('--transcripts', dest='transcripts', help='Transcriptome assembly fasta file')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
23
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
24 args = parser.parse_args()
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
25
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
26 # Build the command line.
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
27 cmd = 'AssemblyPostProcessor'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
28 if args.dereplicate is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
29 cmd += ' --dereplicate'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
30 if args.gap_trimming > 0:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
31 cmd += ' --gap_trimming %4f' % args.gap_trimming
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
32 if args.gene_family_search is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
33 cmd += ' --gene_family_search %s' % args.gene_family_search
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
34 if args.method is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
35 cmd += ' --method %s' % args.method
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
36 if args.min_length > 0:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
37 cmd += ' --min_length %d' % args.min_length
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
38 cmd += ' --num_threads %d' % args.num_threads
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
39 cmd += ' --prediction_method %s' % args.prediction_method
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
40 if args.scaffold is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
41 cmd += ' --scaffold %s' % args.scaffold
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
42 if args.score_matrices is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
43 cmd += ' --score_matrices %s' % args.score_matrices
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
44 if args.strand_specific is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
45 cmd += ' --strand_specific'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
46 cmd += ' --transcripts %s' % args.transcripts
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
47 # Run the command.
49
9d58af3e459d Uploaded
greg
parents: 46
diff changeset
48 utils.run_command(cmd)
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
49
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
50 # Handle outputs.
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
51 if args.output_pttgf is not None and args.output_pttgf_dir is not None:
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
52 src_output_dir = os.path.join(OUTPUT_DIR, 'targeted_gene_families')
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
53 utils.move_directory_files(src_output_dir, args.output_pttgf_dir)
58
e7f264e21d9a Uploaded
greg
parents: 57
diff changeset
54 utils.write_html_output(args.output_pttgf, 'Targeted gene families', args.output_pttgf_dir)