forked from agartland/utils
-
Notifications
You must be signed in to change notification settings - Fork 0
/
parse_immune.py
143 lines (121 loc) · 5.16 KB
/
parse_immune.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
import pandas as pd
import numpy as np
__all__ = ['parseICS',
'parseBAMA',
'parseNAB',
'parseRx',
'unstackIR',
'irLabels',
'icsTicks',
'icsTickLabels']
icsTicks = np.log10([0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1])
icsTickLabels = ['0.01', '0.025', '0.05', '0.1', '0.25', '0.5', '1']
# icsTicks = np.log10([0.01, 0.025, 0.05, 0.1, 0.2, 0.4, 0.6, 0.8, 1])
#icsTickLabels = ['0.01','0.025', '0.05', '0.1','0.2','0.4','0.6','0.8', '1']
def irLabels(c):
poss = ['CD4+', 'CD8+', 'IgG', 'IgA']
for p in poss:
if c.find(p) >=0:
return p
return 'Other'
def unstackIR(df, uVars):
"""Return a response and magnitude df with one row per ptid
and columns for each combination of uVars"""
varFunc = lambda r: ' '.join(r[uVars])
tmpDf = df.copy()
tmpDf['var'] = tmpDf.apply(varFunc, axis=1)
responseDf = tmpDf.pivot(index='ptid', columns='var', values='response')
magDf = tmpDf.pivot(index='ptid', columns='var', values='mag')
return responseDf, magDf
def _parsePTID(v):
"""Returns a string version of a PTID"""
if pd.isnull(v):
out = 'NA'
elif np.isreal(v):
out = '%1.0f' % v
else:
out = v
out = out.replace('-', '')
if out[-2:] == '.0':
out = out[:-2]
return out
def _parseIR(fn, uVars, mag, subset={}, printUnique=False, sep=','):
raw = pd.read_csv(fn, dtype={'ptid':str, 'Ptid':str}, skipinitialspace=True, sep=sep)
raw = raw.rename_axis({'Ptid':'ptid'}, axis=1)
raw.loc[:, 'ptid'] = raw.loc[:, 'ptid'].map(_parsePTID)
allCols = raw.columns.tolist()
if uVars is None:
uVars = raw.columns.drop(['ptid', 'response', mag]).tolist()
if printUnique:
for v in uVars:
u = raw[v].astype(str).unique()
if raw[v].dtype == object or len(u) <= 20:
print('%s: %s' % (v, ', '.join(u)))
else:
print('%s: mean %1.2f' % (v, np.nanmean(raw[v])))
return
cols = []
for c in ['ptid', 'response']:
if c in allCols and not c in uVars:
cols = cols + [c]
cols = cols + uVars + ['mag']
raw['mag'] = raw[mag]
"""Keep rows that have one of the values in v for column k,
for every key/value in subset dict"""
for k, v in list(subset.items()):
raw = raw.loc[raw[k].isin(v)]
ptids = raw['ptid'].unique().shape[0]
total = raw.shape[0]
tmp = raw.set_index(uVars)
conditions = tmp.index.unique().shape[0]
printTuple = (ptids*conditions - total, ptids, conditions, ptids*conditions, total)
if total > (ptids * conditions):
print('uVars are not sufficiently unique (%d): expected %d PTIDs x %d conditions = %d assays, found %d' % printTuple)
elif tmp.shape[0] < (tmp['ptid'].unique().shape[0] * tmp.index.unique().shape[0]):
print('Missing %d assays: expected %d PTIDs x %d conditions = %d assays, found %d' % printTuple)
"""What about dropping the negative controls?"""
return raw[cols]
def parseICS(fn, uVars=['visitno', 'tcellsub', 'cytokine', 'antigen'], mag='pctpos_adj', subset={}, printUnique=False):
"""Parse a processed ICS file.
Returns one row per response, subsetting on subset values."""
out = _parseIR(fn, uVars, mag, subset=subset, printUnique=printUnique)
if not printUnique:
"""Enforce LOD"""
out.loc[out.mag < 0.00025, 'mag'] = 0.00025
out['mag'] = np.log10(out.mag)
return out
def parseBAMA(fn, uVars=['isotype', 'antigen'], mag='delta', subset={}, printUnique=False):
#cols = ['protocol','ptid','antigen','response','delta','rx_code','antigen_label','visitno']
out = _parseIR(fn, uVars, mag, subset=subset, printUnique=printUnique)
if not printUnique:
"""Enforce LOD"""
#out.loc[out.mag < 1, 'mag'] = 1
out['mag'] = np.log10(out.mag)
return out
def parseNAB(fn, uVars=['celltype', 'virusdilution', 'isolate'], mag='titer_num', subset={}, printUnique=False):
out = _parseIR(fn, uVars, mag, subset=subset, printUnique=printUnique)
if not printUnique:
out['mag'] = np.log10(out.mag)
return out
def parseRx(rxFn, demFn=None):
trtCols = ['ptid', 'arm', 'grp', 'protocol', 'rx_code', 'rx']
tmp = pd.read_csv(rxFn)
tmp = tmp.rename_axis({'Ptid': 'ptid'}, axis=1)
tmp.loc[:, 'ptid'] = tmp.ptid.str.replace('-', '')
trtDf = tmp[trtCols].set_index('ptid')
if not demFn is None:
demCols = ['ptid', 'site', 'sex']
demDf = pd.read_csv(demFn)
demDf = demDf.rename_axis({'Ptid': 'ptid'}, axis=1)
demDf.loc[:, 'ptid'] = demDf.ptid.str.replace('-', '')
siteLists = dict(US = [121, 125, 126, 123, 127, 128, 129, 132, 133, 134, 167],
ZA = [138, 156, 157],
Lausanne = [168],
Peru = [150, 621])
siteTranslation = {}
for k, v in list(siteLists.items()):
siteTranslation.update({n:k for n in v})
demDf['site'] = demDf.DEMsitei.map(siteTranslation.get)
demDf['sex'] = demDf.DEMsex
trtDf = trtDf.join(demDf[demCols].set_index('ptid'))
return trtDf