0
|
1 #!/usr/bin/env python
|
|
2 ######################################################################
|
|
3 # Copyright (c) 2016 Northrop Grumman.
|
|
4 # All rights reserved.
|
|
5 ######################################################################
|
|
6 # version 2
|
|
7 from __future__ import print_function
|
|
8
|
|
9 import sys
|
|
10 import os
|
|
11 from argparse import ArgumentParser
|
|
12 import pandas as pd
|
|
13 from scipy.stats import gmean
|
|
14
|
|
15
|
|
16 def run_FLOCK(input_file, method, bins, density, output_file, mfi_file,
|
|
17 mfi_calc, profile, tool_directory):
|
|
18 run_command = tool_directory + "/bin/" + method + " " + input_file
|
|
19 if bins:
|
|
20 run_command += " " + bins
|
|
21 if density:
|
|
22 run_command += " " + density
|
|
23
|
|
24 os.system(run_command)
|
|
25
|
|
26 move_command = "mv flock_results.txt " + output_file
|
|
27 os.system(move_command)
|
|
28
|
|
29 # Here add some way to calculate the count and tack it on to profile file.
|
|
30 flockdf = pd.read_table(output_file)
|
|
31 if mfi_calc == "mfi":
|
|
32 MFIs = flockdf.groupby('Population').mean().round(decimals=2)
|
|
33 elif mfi_calc == "gmfi":
|
|
34 MFIs = flockdf.groupby('Population').agg(lambda x: gmean(list(x))).round(decimals=2)
|
|
35 else:
|
|
36 MFIs = flockdf.groupby('Population').median().round(decimals=2)
|
|
37
|
|
38 with open(mfi_file, "w") as outf:
|
|
39 MFIs.to_csv(outf, sep="\t", float_format='%.0f')
|
|
40
|
|
41 (events, columns) = flockdf.shape
|
|
42 fstats = {}
|
|
43 fstats['population'] = flockdf.iloc[:, -1:].iloc[:, 0]
|
|
44 fstats['population_freq'] = fstats['population'].value_counts()
|
|
45 fstats['population_freq_sort'] = fstats['population_freq'].sort_index()
|
|
46 fstats['population_per'] = (fstats['population'].value_counts(normalize=True) * 100).round(decimals=2)
|
|
47 fstats['population_per_sort'] = fstats['population_per'].sort_index()
|
|
48 fstats['population_all'] = pd.concat([fstats['population_freq_sort'], fstats['population_per_sort']], axis=1)
|
|
49 fstats['population_all'].columns = ['Count', 'Percentage']
|
|
50 fstats['population_all']['Population_ID'] = fstats['population_all'].index
|
|
51
|
|
52 flock_profile = pd.read_table('profile.txt')
|
|
53 profile_pop = flock_profile.merge(fstats['population_all'], on='Population_ID')
|
|
54 profile_pop.to_csv(profile, sep="\t", float_format='%.2f', index=False)
|
|
55
|
|
56 # get_profile = "mv profile.txt " + profile
|
|
57 # os.system(get_profile)
|
|
58 return
|
|
59
|
|
60
|
|
61 if __name__ == "__main__":
|
|
62 parser = ArgumentParser(
|
|
63 prog="runFlockMFI",
|
|
64 description="Run Flock on text file and generate centroid file")
|
|
65
|
|
66 parser.add_argument(
|
|
67 '-i',
|
|
68 dest="input_file",
|
|
69 required=True,
|
|
70 help="File location for the FCS file.")
|
|
71
|
|
72 parser.add_argument(
|
|
73 '-m',
|
|
74 dest="method",
|
|
75 required=True,
|
|
76 help="Run flock1 or flock2.")
|
|
77
|
|
78 parser.add_argument(
|
|
79 '-M',
|
|
80 dest="mfi_calc",
|
|
81 required=True,
|
|
82 help="what to calculate for centroids.")
|
|
83
|
|
84 parser.add_argument(
|
|
85 '-b',
|
|
86 dest="bins",
|
|
87 required=False,
|
|
88 help="Number of Bins.")
|
|
89
|
|
90 parser.add_argument(
|
|
91 '-d',
|
|
92 dest="density",
|
|
93 required=False,
|
|
94 help="Density.")
|
|
95
|
|
96 parser.add_argument(
|
|
97 '-o',
|
|
98 dest="output_file",
|
|
99 required=True,
|
|
100 help="File location for the output file.")
|
|
101
|
|
102 parser.add_argument(
|
|
103 '-t',
|
|
104 dest="tool_directory",
|
|
105 required=True,
|
|
106 help="File location for the output file.")
|
|
107
|
|
108 parser.add_argument(
|
|
109 '-c',
|
|
110 dest="centroids",
|
|
111 required=True,
|
|
112 help="File location for the output centroid file.")
|
|
113
|
|
114 parser.add_argument(
|
|
115 '-p',
|
|
116 dest="profile",
|
|
117 required=True,
|
|
118 help="File location for the output profile file.")
|
|
119
|
|
120 args = parser.parse_args()
|
|
121 run_FLOCK(args.input_file, args.method, args.bins,
|
|
122 args.density, args.output_file, args.centroids, args.mfi_calc,
|
|
123 args.profile, args.tool_directory)
|
|
124
|
|
125 sys.exit(0)
|