|
0
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import optparse
|
|
|
4 import os.path
|
|
|
5
|
|
|
6
|
|
|
7 def fix_header_line(start_header, header_line, new_names):
|
|
|
8 header_parts = header_line.split("\t")
|
|
|
9 if len(header_parts) <= len(start_header):
|
|
|
10 raise Exception("Only found {0} columns in second (header) line expected at least {1}.".format(len(header_parts), (len(start_header) + 1)))
|
|
|
11 data_headers = header_parts[:len(start_header)]
|
|
|
12 if data_headers != start_header:
|
|
|
13 raise Exception("Unexpected start to second (header) line Found: ")
|
|
|
14 new_header = "\t".join(start_header)
|
|
|
15 file_headers = header_parts[len(start_header):]
|
|
|
16 if len(file_headers) != len(new_names):
|
|
|
17 raise Exception("Found {0} file columns in header line, but {1} new_name paramters provided.".format(len(file_headers), len(new_names)))
|
|
|
18 for i in range(len(file_headers)):
|
|
|
19 new_header += "\t"
|
|
|
20 new_header += new_names[i]
|
|
|
21 new_header += "\n"
|
|
|
22 return new_header
|
|
|
23
|
|
|
24
|
|
|
25 def clean_names(prefix, old_names):
|
|
|
26 if len(old_names) > 1:
|
|
|
27 shared_start = old_names[0].strip()
|
|
|
28 shared_ends = old_names[0].strip()
|
|
|
29 for name in old_names:
|
|
|
30 clean = name.strip()
|
|
|
31 while len(shared_start) > 0 and (not clean.startswith(shared_start)):
|
|
|
32 shared_start = shared_start[:-1]
|
|
|
33 while len(shared_ends) > 0 and (not clean.endswith(shared_ends)):
|
|
|
34 shared_ends = shared_ends[1:]
|
|
|
35 start = len(shared_start)
|
|
|
36 end = 0 - len(shared_ends)
|
|
|
37 else:
|
|
|
38 start = 0
|
|
|
39 end = 0
|
|
|
40 new_names = []
|
|
|
41 if end < 0:
|
|
|
42 for name in old_names:
|
|
|
43 new_names.append(prefix + name.strip()[start:end])
|
|
|
44 else:
|
|
|
45 for name in old_names:
|
|
|
46 new_names.append(prefix + name.strip()[start:])
|
|
|
47 return new_names
|
|
|
48
|
|
|
49
|
|
|
50 def main():
|
|
|
51 #Parse Command Line
|
|
|
52 parser = optparse.OptionParser()
|
|
|
53 parser.add_option("--raw_count_file", action="store", type="string", default=None, help="path to file original with the counts")
|
|
|
54 parser.add_option("--fixed_count_file", action="store", type="string", default=None, help="new path for renamaned counts file")
|
|
|
55 parser.add_option("--raw_summary_file", action="store", type="string", default=None, help="path to file original with the summary")
|
|
|
56 parser.add_option("--fixed_summary_file", action="store", type="string", default=None, help="new path for renamaned summary file")
|
|
|
57 parser.add_option("--names_file", action="store", type="string", default=None, help="path to file which contains the names.")
|
|
|
58 parser.add_option("--new_name", action="append", type="string", default=None,
|
|
|
59 help="Names to be used. Must be the same length as in the raw_count_file")
|
|
|
60 parser.add_option("--names_prefix", action="store", type="string", default="", help="Prefix to add in from of every name.")
|
|
|
61
|
|
|
62 (options, args) = parser.parse_args()
|
|
|
63
|
|
|
64 if not os.path.exists(options.raw_count_file):
|
|
|
65 parser.error("Unable to find raw_count_file {0}.".format(options.raw_count_file))
|
|
|
66 if options.names_file:
|
|
|
67 if options.new_name:
|
|
|
68 parser.error("names_file parameter clashes with new_names paramter(s)")
|
|
|
69 if not os.path.exists(options.names_file):
|
|
|
70 parser.error("Unable to find names_file {0}.".format(options.names_file))
|
|
|
71 new_names = []
|
|
|
72 with open(options.names_file, "r") as names_file:
|
|
|
73 for line in names_file:
|
|
|
74 new_names.append(line.strip())
|
|
|
75 new_names = clean_names(options.names_prefix, new_names)
|
|
|
76 else:
|
|
|
77 if not options.new_name:
|
|
|
78 parser.error("No names_file or new_name paraters provided.")
|
|
|
79 new_names = options.new_name
|
|
|
80
|
|
|
81 print "Changing column names to ", new_names
|
|
|
82
|
|
|
83 with open(options.raw_count_file, "r") as input_file:
|
|
|
84 with open(options.fixed_count_file, "w") as output_file:
|
|
|
85 input_file.readline() # job line
|
|
|
86 start_header = ["Geneid", "Chr", "Start", "End", "Strand", "Length"]
|
|
|
87 header_line = fix_header_line(start_header, input_file.readline(), new_names)
|
|
|
88 output_file.write(header_line)
|
|
|
89 for line in input_file:
|
|
2
|
90 output_file.write(line)
|
|
0
|
91
|
|
|
92 with open(options.raw_summary_file, "r") as input_file:
|
|
|
93 with open(options.fixed_summary_file, "w") as output_file:
|
|
|
94 start_header = ["Status"]
|
|
|
95 header_line = fix_header_line(start_header, input_file.readline(), new_names)
|
|
|
96 output_file.write(header_line)
|
|
|
97 for line in input_file:
|
|
|
98 output_file.write(line)
|
|
|
99
|
|
|
100
|
|
|
101 if __name__ == "__main__":
|
|
|
102 main()
|
