annotate assembly_post_processor.py @ 83:ed2cc7e78e9b draft

Uploaded
author greg
date Fri, 13 Oct 2017 14:07:55 -0400
parents 03716365973d
children
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
65
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
4 import shutil
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
5
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
6 import utils
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
7
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
8 OUTPUT_DIR = 'assemblyPostProcessing_dir'
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
9
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
10 parser = argparse.ArgumentParser()
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
11 parser.add_argument('--dereplicate', dest='dereplicate', default=None, help='Remove duplicate sequences')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
12 parser.add_argument('--gap_trimming', dest='gap_trimming', type=float, default=0, help='Trim alignments')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
13 parser.add_argument('--gene_family_search', dest='gene_family_search', default=None, help='Targeted gene families')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
14 parser.add_argument('--method', dest='method', default=None, help='Protein clustering method')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
15 parser.add_argument('--min_length', dest='min_length', type=int, default=0, help='Minimum sequence length')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
16 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of processors')
65
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
17 parser.add_argument('--output_cds', dest='output_cds', help='Output transcripts.cds')
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
18 parser.add_argument('--output_cleaned_cds', dest='output_cleaned_cds', help='Output transcripts.cleaned.cds')
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
19 parser.add_argument('--output_cleaned_nr_cds', dest='output_cleaned_nr_cds', default=None, help='Output transcripts.cleaned.nr.cds')
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
20 parser.add_argument('--output_cleaned_nr_pep', dest='output_cleaned_nr_pep', default=None, help='Output transcripts.cleaned.nr.pep')
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
21 parser.add_argument('--output_cleaned_pep', dest='output_cleaned_pep', help='Output transcripts.cleaned.pep')
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
22 parser.add_argument('--output_pep', dest='output_pep', help='Output transcripts.pep')
83
ed2cc7e78e9b Uploaded
greg
parents: 79
diff changeset
23 parser.add_argument('--output_targeted_gene_families_stats', dest='output_targeted_gene_families_stats', default=None, help='Targeted gene families statistics')
46
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
24 parser.add_argument('--prediction_method', dest='prediction_method', help='Coding regions prediction method')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
25 parser.add_argument('--scaffold', dest='scaffold', default=None, help='Gene family scaffold')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
26 parser.add_argument('--score_matrices', dest='score_matrices', default=None, help='Scores matrices')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
27 parser.add_argument('--strand_specific', dest='strand_specific', default=None, help='Strand-specific assembly')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
28 parser.add_argument('--transcripts', dest='transcripts', help='Transcriptome assembly fasta file')
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
29
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
30 args = parser.parse_args()
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
31
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
32 # Build the command line.
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
33 cmd = 'AssemblyPostProcessor'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
34 if args.dereplicate is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
35 cmd += ' --dereplicate'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
36 if args.gap_trimming > 0:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
37 cmd += ' --gap_trimming %4f' % args.gap_trimming
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
38 if args.gene_family_search is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
39 cmd += ' --gene_family_search %s' % args.gene_family_search
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
40 if args.method is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
41 cmd += ' --method %s' % args.method
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
42 if args.min_length > 0:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
43 cmd += ' --min_length %d' % args.min_length
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
44 cmd += ' --num_threads %d' % args.num_threads
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
45 cmd += ' --prediction_method %s' % args.prediction_method
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
46 if args.scaffold is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
47 cmd += ' --scaffold %s' % args.scaffold
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
48 if args.score_matrices is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
49 cmd += ' --score_matrices %s' % args.score_matrices
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
50 if args.strand_specific is not None:
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
51 cmd += ' --strand_specific'
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
52 cmd += ' --transcripts %s' % args.transcripts
b7858b1f64f1 Uploaded
greg
parents:
diff changeset
53 # Run the command.
49
9d58af3e459d Uploaded
greg
parents: 46
diff changeset
54 utils.run_command(cmd)
54
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
55
9a105e8f08e5 Uploaded
greg
parents: 49
diff changeset
56 # Handle outputs.
65
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
57 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cds'), args.output_cds)
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
58 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.cds'), args.output_cleaned_cds)
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
59 if args.output_cleaned_nr_cds is not None:
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
60 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.nr.cds'), args.output_cleaned_nr_cds)
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
61 if args.output_cleaned_nr_pep is not None:
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
62 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.nr.pep'), args.output_cleaned_nr_pep)
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
63 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.pep'), args.output_cleaned_pep)
baf10f7f3d4a Uploaded
greg
parents: 58
diff changeset
64 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.pep'), args.output_pep)
83
ed2cc7e78e9b Uploaded
greg
parents: 79
diff changeset
65 if args.output_targeted_gene_families_stats is not None:
ed2cc7e78e9b Uploaded
greg
parents: 79
diff changeset
66 shutil.move(os.path.join(OUTPUT_DIR, 'targeted_gene_family_assemblies.stats'), args.output_targeted_gene_families_stats)