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