comparison auto_collapse_pop/auto_collapse_pops.py @ 0:7632562ada41 draft

Uploaded
author immport-devteam
date Mon, 27 Feb 2017 12:14:29 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:7632562ada41
1 #!/usr/bin/env python
2
3 ######################################################################
4 # Copyright (c) 2016 Northrop Grumman.
5 # All rights reserved.
6 ######################################################################
7 from __future__ import print_function
8 import sys
9 import pandas as pd
10 from argparse import ArgumentParser
11
12
13 def auto_collapse(input_file, profile_file, output, report):
14 profile_pop_list = {}
15 pop_to_collapse = []
16 markers = []
17 with open(profile_file, "r") as pf:
18 pffl = pf.readline()
19 markers = pffl.strip().split("\t")
20 for pfline in pf:
21 line = pfline.strip().split("\t")
22 pop = line[0]
23 profil = "\t".join(line[1:-2])
24 if profil in profile_pop_list:
25 profile_pop_list[profil].append(pop)
26 else:
27 profile_pop_list[profil] = [pop]
28 i = 1
29 with open(report, "w") as rt:
30 rt.write("New_Population\tFormer_Populations\t")
31 rt.write("\t".join(markers[1:-2]) + "\n")
32 for profs in profile_pop_list:
33 pop_to_collapse.append(profile_pop_list[profs])
34 pop_ls = ", ".join(profile_pop_list[profs])
35 rt.write("\t".join([str(i), pop_ls, profs]) + "\n")
36 i += 1
37 df = pd.read_table(input_file, dtype={'Population': object})
38 df['new_population'] = df.Population
39 for i, sets_pop in enumerate(pop_to_collapse):
40 df.loc[df['Population'].isin(sets_pop), ['new_population']] = i + 1
41
42 df.Population = df.new_population
43 df.drop(['new_population'], inplace=True, axis=1)
44 df.to_csv(output, sep="\t", index=False)
45
46
47 if __name__ == "__main__":
48 parser = ArgumentParser(
49 prog="auto_pop_collapse_from_profile",
50 description="collapses FLOCK populations based on profile.")
51
52 parser.add_argument(
53 '-i',
54 dest="input_file",
55 required=True,
56 help="FLOCK output file")
57
58 parser.add_argument(
59 '-o',
60 dest="output",
61 required=True,
62 help="Name of the output file.")
63
64 parser.add_argument(
65 '-r',
66 dest="report",
67 required=True,
68 help="Name of the report file.")
69
70 parser.add_argument(
71 '-p',
72 dest="profile",
73 required=True,
74 help="File location for the profile.txt from FLOCK.")
75
76 args = parser.parse_args()
77
78 auto_collapse(args.input_file, args.profile, args.output, args.report)
79 sys.exit(0)