annotate convert_bc_to_binary_RY.py @ 37:744e4f2b972c draft

try pybedtools 0.6.9
author rnateam
date Fri, 18 Dec 2015 03:49:54 -0500
parents 17ef0e0dae68
children 0b9aab6aaebf
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
1 #!/usr/bin/env python
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
3 tool_description = """
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
4 Convert standard nucleotides to IUPAC nucleotide codes used for binary barcodes.
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
5
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
6 A and G are converted to nucleotide code R. T, U and C are converted to Y. By
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
7 default output is written to stdout.
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
8
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
9 Example usage:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
10 - write converted sequences from file in.fa to file file out.fa:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
11 convert_bc_to_binary_RY.py in.fa --outfile out.fa
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
12 """
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
13
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
14 epilog = """
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
15 Author: Daniel Maticzka
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
16 Copyright: 2015
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
17 License: Apache
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
18 Email: maticzkd@informatik.uni-freiburg.de
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
19 Status: Testing
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
20 """
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
21
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
22 import argparse
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
23 import logging
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
24 from string import maketrans
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
25 from sys import stdout
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
26 from Bio import SeqIO
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
27 from Bio.Seq import Seq
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
28 from Bio.Alphabet import IUPAC
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
29
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
30 # # avoid ugly python IOError when stdout output is piped into another program
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
31 # # and then truncated (such as piping to head)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
32 # from signal import signal, SIGPIPE, SIG_DFL
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
33 # signal(SIGPIPE, SIG_DFL)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
34
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
35 # parse command line arguments
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
36 parser = argparse.ArgumentParser(description=tool_description,
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
37 epilog=epilog,
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
38 formatter_class=argparse.RawDescriptionHelpFormatter)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
39 # positional arguments
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
40 parser.add_argument(
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
41 "infile",
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
42 help="Path to fastq input file.")
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
43 # optional arguments
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
44 parser.add_argument(
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
45 "-o", "--outfile",
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
46 help="Write results to this file.")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
47 parser.add_argument(
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
48 "--fasta-format",
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
49 dest="fasta_format",
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
50 help="Read and write fasta instead of fastq format.",
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
51 action="store_true")
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
52 parser.add_argument(
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
53 "-v", "--verbose",
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
54 help="Be verbose.",
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
55 action="store_true")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
56 parser.add_argument(
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
57 "-d", "--debug",
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
58 help="Print lots of debugging information",
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
59 action="store_true")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
60 parser.add_argument(
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
61 '--version',
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
62 action='version',
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
63 version='0.1.0')
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
64
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
65
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
66 def translate_nt_to_RY(seq):
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
67 """Translates nucleotides to RY (A,G -> R; C,U,T -> Y).
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
68
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
69 >>> translate_nt_to_RY("ACGUTACGUT")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
70 RYRYYRYRYY
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
71 """
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
72 trans_table = maketrans("AGCUT", "RRYYY")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
73 trans_seq = seq.translate(trans_table)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
74 logging.debug(seq + " -> " + trans_seq)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
75 return trans_seq
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
76
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
77
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
78 def translate_nt_to_RY_iterator(robj):
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
79 """Translate SeqRecords sequences to RY alphabet."""
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
80 for record in robj:
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
81 if not args.fasta_format:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
82 saved_letter_annotations = record.letter_annotations
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
83 record.letter_annotations = {}
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
84 record.seq = Seq(translate_nt_to_RY(str(record.seq)),
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
85 IUPAC.unambiguous_dna)
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
86 if not args.fasta_format:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
87 record.letter_annotations = saved_letter_annotations
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
88 yield record
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
89
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
90 # handle arguments
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
91 args = parser.parse_args()
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
92 if args.debug:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
93 logging.basicConfig(level=logging.DEBUG, format="%(asctime)s - %(filename)s - %(levelname)s - %(message)s")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
94 elif args.verbose:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
95 logging.basicConfig(level=logging.INFO, format="%(filename)s - %(levelname)s - %(message)s")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
96 else:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
97 logging.basicConfig(format="%(filename)s - %(levelname)s - %(message)s")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
98 logging.info("Parsed arguments:")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
99 if args.outfile:
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
100 logging.info(" outfile: enabled writing to file")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
101 logging.info(" outfile: '{}'".format(args.outfile))
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
102 logging.info(" outfile: '{}'".format(args.outfile))
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
103 logging.info("")
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
104
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
105 # get input iterator
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
106 input_handle = open(args.infile, "rU")
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
107 if args.fasta_format:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
108 input_seq_iterator = SeqIO.parse(input_handle, "fasta")
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
109 else:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
110 input_seq_iterator = SeqIO.parse(input_handle, "fastq")
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
111 convert_seq_iterator = translate_nt_to_RY_iterator(input_seq_iterator)
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
112 output_handle = (open(args.outfile, "w") if args.outfile is not None else stdout)
8
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
113 if args.fasta_format:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
114 SeqIO.write(convert_seq_iterator, output_handle, "fasta")
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
115 else:
17ef0e0dae68 Uploaded
rnateam
parents: 2
diff changeset
116 SeqIO.write(convert_seq_iterator, output_handle, "fastq")
2
de4ea3aa1090 Uploaded
rnateam
parents:
diff changeset
117 output_handle.close()