annotate assembly_post_processor.py @ 3:07ee186ba16e draft

Uploaded
author greg
date Thu, 24 Aug 2017 13:10:29 -0400
parents fcc558568020
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
fcc558568020 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env python
fcc558568020 Uploaded
greg
parents:
diff changeset
2 import argparse
fcc558568020 Uploaded
greg
parents:
diff changeset
3 import os
fcc558568020 Uploaded
greg
parents:
diff changeset
4 import shutil
fcc558568020 Uploaded
greg
parents:
diff changeset
5
fcc558568020 Uploaded
greg
parents:
diff changeset
6 import utils
fcc558568020 Uploaded
greg
parents:
diff changeset
7
fcc558568020 Uploaded
greg
parents:
diff changeset
8 OUTPUT_DIR = 'assemblyPostProcessing_dir'
fcc558568020 Uploaded
greg
parents:
diff changeset
9
fcc558568020 Uploaded
greg
parents:
diff changeset
10 parser = argparse.ArgumentParser()
fcc558568020 Uploaded
greg
parents:
diff changeset
11 parser.add_argument('--dereplicate', dest='dereplicate', default=None, help='Remove duplicate sequences')
fcc558568020 Uploaded
greg
parents:
diff changeset
12 parser.add_argument('--gap_trimming', dest='gap_trimming', type=float, default=0, help='Trim alignments')
fcc558568020 Uploaded
greg
parents:
diff changeset
13 parser.add_argument('--gene_family_search', dest='gene_family_search', default=None, help='Targeted gene families')
fcc558568020 Uploaded
greg
parents:
diff changeset
14 parser.add_argument('--method', dest='method', default=None, help='Protein clustering method')
fcc558568020 Uploaded
greg
parents:
diff changeset
15 parser.add_argument('--min_length', dest='min_length', type=int, default=0, help='Minimum sequence length')
fcc558568020 Uploaded
greg
parents:
diff changeset
16 parser.add_argument('--num_threads', dest='num_threads', type=int, help='Number of processors')
fcc558568020 Uploaded
greg
parents:
diff changeset
17 parser.add_argument('--output_pttgf', dest='output_pttgf', default=None, help='Primary targeted gene families dataset')
fcc558568020 Uploaded
greg
parents:
diff changeset
18 parser.add_argument('--output_cds', dest='output_cds', help='Output transcripts.cds')
fcc558568020 Uploaded
greg
parents:
diff changeset
19 parser.add_argument('--output_cleaned_cds', dest='output_cleaned_cds', help='Output transcripts.cleaned.cds')
fcc558568020 Uploaded
greg
parents:
diff changeset
20 parser.add_argument('--output_cleaned_nr_cds', dest='output_cleaned_nr_cds', default=None, help='Output transcripts.cleaned.nr.cds')
fcc558568020 Uploaded
greg
parents:
diff changeset
21 parser.add_argument('--output_cleaned_nr_pep', dest='output_cleaned_nr_pep', default=None, help='Output transcripts.cleaned.nr.pep')
fcc558568020 Uploaded
greg
parents:
diff changeset
22 parser.add_argument('--output_cleaned_pep', dest='output_cleaned_pep', help='Output transcripts.cleaned.pep')
fcc558568020 Uploaded
greg
parents:
diff changeset
23 parser.add_argument('--output_pep', dest='output_pep', help='Output transcripts.pep')
fcc558568020 Uploaded
greg
parents:
diff changeset
24 parser.add_argument('--output_pttgf_dir', dest='output_pttgf_dir', default=None, help='Directory hierarchy of targeted gene family datasets')
fcc558568020 Uploaded
greg
parents:
diff changeset
25 parser.add_argument('--prediction_method', dest='prediction_method', help='Coding regions prediction method')
fcc558568020 Uploaded
greg
parents:
diff changeset
26 parser.add_argument('--scaffold', dest='scaffold', default=None, help='Gene family scaffold')
fcc558568020 Uploaded
greg
parents:
diff changeset
27 parser.add_argument('--score_matrices', dest='score_matrices', default=None, help='Scores matrices')
fcc558568020 Uploaded
greg
parents:
diff changeset
28 parser.add_argument('--strand_specific', dest='strand_specific', default=None, help='Strand-specific assembly')
fcc558568020 Uploaded
greg
parents:
diff changeset
29 parser.add_argument('--transcripts', dest='transcripts', help='Transcriptome assembly fasta file')
fcc558568020 Uploaded
greg
parents:
diff changeset
30
fcc558568020 Uploaded
greg
parents:
diff changeset
31 args = parser.parse_args()
fcc558568020 Uploaded
greg
parents:
diff changeset
32
fcc558568020 Uploaded
greg
parents:
diff changeset
33 # Build the command line.
fcc558568020 Uploaded
greg
parents:
diff changeset
34 cmd = 'AssemblyPostProcessor'
fcc558568020 Uploaded
greg
parents:
diff changeset
35 if args.dereplicate is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
36 cmd += ' --dereplicate'
fcc558568020 Uploaded
greg
parents:
diff changeset
37 if args.gap_trimming > 0:
fcc558568020 Uploaded
greg
parents:
diff changeset
38 cmd += ' --gap_trimming %4f' % args.gap_trimming
fcc558568020 Uploaded
greg
parents:
diff changeset
39 if args.gene_family_search is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
40 cmd += ' --gene_family_search %s' % args.gene_family_search
fcc558568020 Uploaded
greg
parents:
diff changeset
41 if args.method is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
42 cmd += ' --method %s' % args.method
fcc558568020 Uploaded
greg
parents:
diff changeset
43 if args.min_length > 0:
fcc558568020 Uploaded
greg
parents:
diff changeset
44 cmd += ' --min_length %d' % args.min_length
fcc558568020 Uploaded
greg
parents:
diff changeset
45 cmd += ' --num_threads %d' % args.num_threads
fcc558568020 Uploaded
greg
parents:
diff changeset
46 cmd += ' --prediction_method %s' % args.prediction_method
fcc558568020 Uploaded
greg
parents:
diff changeset
47 if args.scaffold is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
48 cmd += ' --scaffold %s' % args.scaffold
fcc558568020 Uploaded
greg
parents:
diff changeset
49 if args.score_matrices is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
50 cmd += ' --score_matrices %s' % args.score_matrices
fcc558568020 Uploaded
greg
parents:
diff changeset
51 if args.strand_specific is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
52 cmd += ' --strand_specific'
fcc558568020 Uploaded
greg
parents:
diff changeset
53 cmd += ' --transcripts %s' % args.transcripts
fcc558568020 Uploaded
greg
parents:
diff changeset
54 # Run the command.
fcc558568020 Uploaded
greg
parents:
diff changeset
55 utils.run_command(cmd)
fcc558568020 Uploaded
greg
parents:
diff changeset
56
fcc558568020 Uploaded
greg
parents:
diff changeset
57 # Handle outputs.
fcc558568020 Uploaded
greg
parents:
diff changeset
58 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cds'), args.output_cds)
fcc558568020 Uploaded
greg
parents:
diff changeset
59 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.cds'), args.output_cleaned_cds)
fcc558568020 Uploaded
greg
parents:
diff changeset
60 if args.output_cleaned_nr_cds is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
61 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.nr.cds'), args.output_cleaned_nr_cds)
fcc558568020 Uploaded
greg
parents:
diff changeset
62 if args.output_cleaned_nr_pep is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
63 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.nr.pep'), args.output_cleaned_nr_pep)
fcc558568020 Uploaded
greg
parents:
diff changeset
64 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.cleaned.pep'), args.output_cleaned_pep)
fcc558568020 Uploaded
greg
parents:
diff changeset
65 shutil.move(os.path.join(OUTPUT_DIR, 'transcripts.pep'), args.output_pep)
fcc558568020 Uploaded
greg
parents:
diff changeset
66 if args.output_pttgf is not None and args.output_pttgf_dir is not None:
fcc558568020 Uploaded
greg
parents:
diff changeset
67 src_output_dir = os.path.join(OUTPUT_DIR, 'targeted_gene_families')
fcc558568020 Uploaded
greg
parents:
diff changeset
68 utils.move_directory_files(src_output_dir, args.output_pttgf_dir)
fcc558568020 Uploaded
greg
parents:
diff changeset
69 utils.write_html_output(args.output_pttgf, 'Targeted gene families', args.output_pttgf_dir)