Mercurial > repos > artbio > repenrich

annotate RepEnrich_setup.py @ 0:f6f0f1e5e940 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
author artbio
date Wed, 02 Aug 2017 05:17:29 -0400
parents
children 6bba3e33c2e7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
1 #!/usr/bin/env python
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
2 import argparse
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
3 import csv
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
4 import os
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
5 import shlex
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
6 import subprocess
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
7 import sys
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
8
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
9 from Bio import SeqIO
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
10 from Bio.Alphabet import IUPAC
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
11 from Bio.Seq import Seq
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
12 from Bio.SeqRecord import SeqRecord
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
13
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
14 parser = argparse.ArgumentParser(description='''
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
15 Part I: Prepartion of repetive element psuedogenomes and repetive\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
16 element bamfiles. This script prepares the annotation used by\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
17 downstream applications to analyze for repetitive element\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
18 enrichment. For this script to run properly bowtie must be\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
19 loaded. The repeat element psuedogenomes are prepared in order\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
20 to analyze reads that map to multiple locations of the genome.\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
21 The repeat element bamfiles are prepared in order to use a\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
22 region sorter to analyze reads that map to a single location\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
23 of the genome. You will 1) annotation_file:\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
24 The repetitive element annotation file downloaded from\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
25 RepeatMasker.org database for your organism of interest.\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
26 2) genomefasta: Your genome of interest in fasta format,\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
27 3)setup_folder: a folder to contain repeat element setup files\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
28 command-line usage
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
29 EXAMPLE: python master_setup.py\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
30 /users/nneretti/data/annotation/mm9/mm9_repeatmasker.txt\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
31 /users/nneretti/data/annotation/mm9/mm9.fa\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
32 /users/nneretti/data/annotation/mm9/setup_folder''',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
33 prog='getargs_genome_maker.py')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
34 parser.add_argument('--version', action='version', version='%(prog)s 0.1')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
35 parser.add_argument('annotation_file', action='store',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
36 metavar='annotation_file',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
37 help='''List annotation file. The annotation file contains\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
38 the repeat masker annotation for the genome of\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
39 interest and may be downloaded at RepeatMasker.org\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
40 Example /data/annotation/mm9/mm9.fa.out''')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
41 parser.add_argument('genomefasta', action='store', metavar='genomefasta',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
42 help='''File name and path for genome of interest in fasta\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
43 format. Example /data/annotation/mm9/mm9.fa''')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
44 parser.add_argument('setup_folder', action='store', metavar='setup_folder',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
45 help='''List folder to contain bamfiles for repeats and\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
46 repeat element psuedogenomes.\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
47 Example /data/annotation/mm9/setup''')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
48 parser.add_argument('--nfragmentsfile1', action='store',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
49 dest='nfragmentsfile1', metavar='nfragmentsfile1',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
50 default='./repnames_nfragments.txt',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
51 help='''Output location of a description file that saves\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
52 the number of fragments processed per repname.
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
53 Default ./repnames_nfragments.txt''')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
54 parser.add_argument('--gaplength', action='store', dest='gaplength',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
55 metavar='gaplength', default='200', type=int,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
56 help='Length of the spacer used to build\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
57 repeat psuedogeneomes. Default 200')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
58 parser.add_argument('--flankinglength', action='store', dest='flankinglength',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
59 metavar='flankinglength', default='25', type=int,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
60 help='Length of the flanking region adjacent to the repeat\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
61 element that is used to build repeat psuedogeneomes.\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
62 The flanking length should be set according to the\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
63 length of your reads. Default 25')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
64 parser.add_argument('--is_bed', action='store', dest='is_bed',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
65 metavar='is_bed', default='FALSE',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
66 help='''Is the annotation file a bed file. This is also a\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
67 compatible format. The file needs to be a tab\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
68 separated bed with optional fields.
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
69 Ex. format:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
70 chr\tstart\tend\tName_element\tclass\tfamily.
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
71 The class and family should identical to name_element\
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
72 if not applicable. Default FALSE change to TRUE''')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
73 args = parser.parse_args()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
74
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
75 # parameters and paths specified in args_parse
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
76 gapl = args.gaplength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
77 flankingl = args.flankinglength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
78 annotation_file = args.annotation_file
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
79 genomefasta = args.genomefasta
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
80 setup_folder = args.setup_folder
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
81 nfragmentsfile1 = args.nfragmentsfile1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
82 is_bed = args.is_bed
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
83
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
84 ##############################################################################
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
85 # check that the programs we need are available
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
86 try:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
87 subprocess.call(shlex.split("bowtie --version"),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
88 stdout=open(os.devnull, 'wb'),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
89 stderr=open(os.devnull, 'wb'))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
90 except OSError:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
91 print("Error: Bowtie or BEDTools not loaded")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
92 raise
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
93
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
94 ##############################################################################
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
95 # Define a text importer
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
96 csv.field_size_limit(sys.maxsize)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
97
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
98
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
99 def import_text(filename, separator):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
100 for line in csv.reader(open(os.path.realpath(filename)),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
101 delimiter=separator, skipinitialspace=True):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
102 if line:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
103 yield line
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
104
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
105
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
106 # Make a setup folder
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
107 if not os.path.exists(setup_folder):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
108 os.makedirs(setup_folder)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
109 ##############################################################################
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
110 # load genome into dictionary
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
111 print("loading genome...")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
112 g = SeqIO.to_dict(SeqIO.parse(genomefasta, "fasta"))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
113
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
114 print("Precomputing length of all chromosomes...")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
115 idxgenome = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
116 lgenome = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
117 genome = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
118 allchrs = g.keys()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
119 k = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
120 for chr in allchrs:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
121 genome[chr] = str(g[chr].seq)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
122 # del g[chr]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
123 lgenome[chr] = len(genome[chr])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
124 idxgenome[chr] = k
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
125 k = k + 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
126 del g
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
127
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
128 ##############################################################################
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
129 # Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
130 if is_bed == "FALSE":
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
131 repeat_elements = []
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
132 fout = open(os.path.realpath(setup_folder + os.path.sep
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
133 + 'repnames.bed'), 'w')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
134 fin = import_text(annotation_file, ' ')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
135 x = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
136 rep_chr = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
137 rep_start = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
138 rep_end = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
139 x = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
140 for line in fin:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
141 if x > 2:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
142 line9 = line[9].replace("(", "_").replace(")",
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
143 "_").replace("/", "_")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
144 repname = line9
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
145 if repname not in repeat_elements:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
146 repeat_elements.append(repname)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
147 repchr = line[4]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
148 repstart = int(line[5])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
149 repend = int(line[6])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
150 fout.write(str(repchr) + '\t' + str(repstart) + '\t' + str(repend)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
151 + '\t' + str(repname) + '\n')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
152 if repname in rep_chr:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
153 rep_chr[repname].append(repchr)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
154 rep_start[repname].append(int(repstart))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
155 rep_end[repname].append(int(repend))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
156 else:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
157 rep_chr[repname] = [repchr]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
158 rep_start[repname] = [int(repstart)]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
159 rep_end[repname] = [int(repend)]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
160 x += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
161 if is_bed == "TRUE":
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
162 repeat_elements = []
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
163 fout = open(os.path.realpath(setup_folder + os.path.sep + 'repnames.bed'),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
164 'w')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
165 fin = open(os.path.realpath(annotation_file), 'r')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
166 x = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
167 rep_chr = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
168 rep_start = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
169 rep_end = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
170 x = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
171 for line in fin:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
172 line = line.strip('\n')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
173 line = line.split('\t')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
174 line3 = line[3].replace("(", "_").replace(")", "_").replace("/", "_")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
175 repname = line3
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
176 if repname not in repeat_elements:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
177 repeat_elements.append(repname)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
178 repchr = line[0]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
179 repstart = int(line[1])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
180 repend = int(line[2])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
181 fout.write(str(repchr) + '\t' + str(repstart) + '\t' +
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
182 str(repend) + '\t' + str(repname) + '\n')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
183 # if rep_chr.has_key(repname):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
184 if repname in rep_chr:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
185 rep_chr[repname].append(repchr)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
186 rep_start[repname].append(int(repstart))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
187 rep_end[repname].append(int(repend))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
188 else:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
189 rep_chr[repname] = [repchr]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
190 rep_start[repname] = [int(repstart)]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
191 rep_end[repname] = [int(repend)]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
192
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
193 fin.close()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
194 fout.close()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
195 repeat_elements = sorted(repeat_elements)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
196 print("Writing a key for all repeats...")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
197 # print to fout the binary key that contains each repeat type with the
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
198 # associated binary number; sort the binary key:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
199 fout = open(os.path.realpath(setup_folder + os.path.sep +
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
200 'repgenomes_key.txt'), 'w')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
201 x = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
202 for repeat in repeat_elements:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
203 # print >> fout, str(repeat) + '\t' + str(x)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
204 fout.write(str(repeat) + '\t' + str(x) + '\n')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
205 x += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
206 fout.close()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
207 ##############################################################################
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
208 # generate spacer for psuedogenomes
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
209 spacer = ""
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
210 for i in range(gapl):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
211 spacer = spacer + "N"
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
212
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
213 # save file with number of fragments processed per repname
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
214 print("Saving number of fragments processed per repname to "
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
215 + nfragmentsfile1)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
216 fout1 = open(os.path.realpath(nfragmentsfile1), "w")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
217 for repname in rep_chr.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
218 rep_chr_current = rep_chr[repname]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
219 # print >>fout1, str(len(rep_chr[repname])) + "\t" + repname
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
220 fout1.write(str(len(rep_chr[repname])) + "\t" + repname + '\n')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
221 fout1.close()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
222
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
223 # generate metagenomes and save them to FASTA files
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
224 k = 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
225 nrepgenomes = len(rep_chr.keys())
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
226 for repname in rep_chr.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
227 metagenome = ""
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
228 newname = repname.replace("(", "_").replace(")", "_").replace("/", "_")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
229 print("processing repgenome " + newname + ".fa" + " (" + str(k)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
230 + " of " + str(nrepgenomes) + ")")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
231 rep_chr_current = rep_chr[repname]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
232 rep_start_current = rep_start[repname]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
233 rep_end_current = rep_end[repname]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
234 print("-------> " + str(len(rep_chr[repname])) + " fragments")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
235 for i in range(len(rep_chr[repname])):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
236 try:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
237 chr = rep_chr_current[i]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
238 rstart = max(rep_start_current[i] - flankingl, 0)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
239 rend = min(rep_end_current[i] + flankingl, lgenome[chr]-1)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
240 metagenome = metagenome + spacer + genome[chr][rstart:(rend+1)]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
241 except KeyError:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
242 print("Unrecognised Chromosome: "+chr)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
243 pass
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
244 # Convert metagenome to SeqRecord object (required by SeqIO.write)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
245 record = SeqRecord(Seq(metagenome, IUPAC.unambiguous_dna), id="repname",
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
246 name="", description="")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
247 print("saving repgenome " + newname + ".fa" + " (" + str(k) + " of "
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
248 + str(nrepgenomes) + ")")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
249 fastafilename = os.path.realpath(setup_folder + os.path.sep
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
250 + newname + ".fa")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
251 SeqIO.write(record, fastafilename, "fasta")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
252 print("indexing repgenome " + newname + ".fa" + " (" +
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
253 str(k) + " of " + str(nrepgenomes) + ")")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
254 command = shlex.split('bowtie-build -f ' + fastafilename + ' ' +
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
255 setup_folder + os.path.sep + newname)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
256 p = subprocess.Popen(command).communicate()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
257 k += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
258
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
259 print("... Done")