forked from iDrDex/star_api
-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.py
287 lines (232 loc) · 9.84 KB
/
main.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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
__author__ = 'dex', 'dhimmel'
import sys
import os
import shutil
import urllib2
import StringIO
import gzip
import re
from funcy import cat, first, re_all
import psycopg2
import psycopg2.extras
import pandas as pd
import numpy as np
import conf
###connect to DB###
import db_conf #PRIVATE
conn = psycopg2.connect(db_conf.DB_PARAMATERS)
cursor = conn.cursor(cursor_factory=psycopg2.extras.DictCursor)
def __getMatrixNumHeaderLines(inStream):
import re
p = re.compile(r'^"ID_REF"')
for i, line in enumerate(inStream):
if p.search(line):
return i
def matrix_filenames(series_id, platform_id):
gse_name = query_record(series_id, "series")['gse_name']
yield "%s/%s_series_matrix.txt.gz" % (gse_name, gse_name)
gpl_name = query_record(platform_id, "platform")['gpl_name']
yield "%s/%s-%s_series_matrix.txt.gz" % (gse_name, gse_name, gpl_name)
def get_matrix_filename(series_id, platform_id):
filenames = list(matrix_filenames(series_id, platform_id))
mirror_filenames = (os.path.join(conf.SERIES_MATRIX_MIRROR, filename) for filename in filenames)
mirror_filename = first(filename for filename in mirror_filenames if os.path.isfile(filename))
if mirror_filename:
return mirror_filename
for filename in filenames:
print 'Loading URL', conf.SERIES_MATRIX_URL + filename, '...'
try:
res = urllib2.urlopen(conf.SERIES_MATRIX_URL + filename)
except urllib2.URLError:
pass
else:
mirror_filename = os.path.join(conf.SERIES_MATRIX_MIRROR, filename)
print 'Cache to', mirror_filename
directory = os.path.dirname(mirror_filename)
if not os.path.exists(directory):
os.makedirs(directory)
with open(mirror_filename, 'wb') as f:
shutil.copyfileobj(res, f)
return mirror_filename
raise LookupError("Can't find matrix file for series %s, platform %s"
% (series_id, platform_id))
def get_data(series_id, platform_id, impute = False):
matrixFilename = get_matrix_filename(series_id, platform_id)
# setup data for specific platform
for attempt in (0, 1):
try:
headerRows = __getMatrixNumHeaderLines(gzip.open(matrixFilename))
na_values = ["null", "NA", "NaN", "N/A", "na", "n/a", ""]
data = pd.io.parsers.read_table(gzip.open(matrixFilename),
skiprows=headerRows,
index_col=["ID_REF"],
na_values=na_values,
skipfooter=1,
engine='python')
except IOError as e:
# In case we have corrupt file
print "Failed loading %s: %s" % (matrixFilename, e)
os.remove(matrixFilename)
if attempt:
raise
matrixFilename = get_matrix_filename(series_id, platform_id)
data = clean_data(data) #drop samples
if len(data.columns) == 1:
data = data.dropna()
elif impute:
data = impute_data(data)
data = log_data(data) #logc
data.index = data.index.astype(str)
data.index.name = "probe"
data.columns.name = 'gsm_name'
for column in data.columns:
data[column] = data[column].astype(np.float64)
# data.to_csv("float64.data.csv")
return data
def get_platform_probes(platform_id):
sql = "select * from platform_probe where platform_id = %s"
return pd.read_sql(sql, conn, "probe", params=(platform_id,))
def query_platform_probes(gpl_name):
platform_id = query_record(gpl_name, "platform", "gpl_name")['id']
return get_platform_probes(platform_id)
def get_samples(series_id, platform_id):
sql = "select * from sample where series_id = %s and platform_id = %s"
return pd.read_sql(sql, conn, "id", params=(series_id, platform_id,))
def query_samples(gse_name, gpl_name):
series_id = query_record(gse_name, "series", "gse_name")['id']
platform_id = query_record(gpl_name, "platform", "gpl_name")['id']
return get_samples(series_id, platform_id)
def get_gene_data(series_id, platform_id):
data = get_data(series_id, platform_id)
platform_probes = get_platform_probes(platform_id)
gene_data = platform_probes[['mygene_sym', 'mygene_entrez']] \
.join(data) \
.set_index(['mygene_sym', 'mygene_entrez'])
gene_data.columns.name = 'gsm_name'
return gene_data
def query_record(id, table, id_field="id"):
sql = """select * from %s where %s """ % (table, id_field) + """= %s"""
cursor.execute(sql, (id,))
return cursor.fetchone()
def query_gene_data(gse_name, gpl_name):
series_id = query_record(gse_name, "series", "gse_name")['id']
platform_id = query_record(gpl_name, "platform", "gpl_name")['id']
gene_data = get_gene_data(series_id, platform_id)
gene_data.columns = gene_data.columns + "_" + gpl_name + "_" + gse_name
return gene_data
def query_data(gse_name, gpl_name, impute=False):
series_id = query_record(gse_name, "series", "gse_name")['id']
platform_id = query_record(gpl_name, "platform", "gpl_name")['id']
data = get_data(series_id, platform_id, impute)
# data.columns = data.columns + "_" + gpl_name + "_" + gse_name
return data
def query_tags_annotations(tokens):
df = pd.read_sql('''
SELECT
sample_id,
sample.gsm_name,
annotation,
series_annotation.series_id,
series.gse_name,
series_annotation.platform_id,
platform.gpl_name,
tag.tag_name
FROM
sample_annotation
JOIN sample ON (sample_annotation.sample_id = sample.id)
JOIN series_annotation ON (sample_annotation.serie_annotation_id = series_annotation.id)
JOIN platform ON (series_annotation.platform_id = platform.id)
JOIN tag ON (series_annotation.tag_id = tag.id)
JOIN series ON (series_annotation.series_id = series.id)
WHERE
tag.tag_name ~* %(tags)s
''', conn, params={'tags': '^(%s)$' % '|'.join(map(re.escape, tokens))})
# wide = get_wide_annotations(df, tokens)
return df
def get_unique_annotations(df):
df= df.query(""" annotation != ''""")\
.groupby(['sample_id', 'series_id', 'platform_id', 'gsm_name', 'gpl_name'],
as_index=False)\
.filter(lambda x: len(x) == 1)
return get_wide_annotations(df)
def get_wide_annotations(df):
tokens = df.tag_name.unique()
# df = df.groupby(['sample_id', 'series_id', 'platform_id', 'gsm_name', 'gpl_name'],
# as_index=False).filter(lambda x: len(x) == 1) #extracts unique
# Make tag columns
df.tag_name = df.tag_name.str.lower()
df.annotation = df.annotation.str.lower()
# create outcome column
# df['outcome'] = None
for tag in tokens:
tag_name = tag.lower()
df[tag_name] = df[df.tag_name == tag_name].annotation
return df
def get_annotations(case_query, control_query, modifier_query=""):
# Fetch all relevant data
queries = [case_query, control_query, modifier_query]
tokens = set(cat(re_all('[a-zA-Z]\w*', query) for query in queries))
df = query_tags_annotations(tokens)
# Make tag columns
df.tag_name = df.tag_name.str.lower()
df.annotation = df.annotation.str.lower()
for tag in tokens:
tag_name = tag.lower()
df[tag_name] = df[df.tag_name == tag_name].annotation
# Select only cells with filled annotations
df = df.drop(['tag_name', 'annotation'], axis=1)
df = df.groupby(['sample_id', 'series_id', 'platform_id', 'gsm_name', 'gpl_name'],
as_index=False).first()
df = df.convert_objects(convert_numeric=True)
# Apply case/control/modifier
if modifier_query:
df = df.query(modifier_query.lower())
case_df = df.query(case_query.lower())
control_df = df.query(control_query.lower())
# Set 0 and 1 for analysis
overlap_df = df.ix[set(case_df.index).intersection(set(control_df.index))]
df['sample_class'] = None
df['sample_class'].ix[case_df.index] = 1
df['sample_class'].ix[control_df.index] = 0
df['sample_class'].ix[overlap_df.index] = -1
return df.dropna(subset=["sample_class"])
import numexpr as ne
def log_data(df):
if is_logged(df):
return df
data = df.values
floor = np.abs(np.nanmin(data, axis=0))
res = ne.evaluate('log(data + floor + 1) / log(2)')
return pd.DataFrame(res, index=df.index, columns=df.columns)
def is_logged(df):
return np.max(df.values) < 10
def impute_data(data):
import rpy2.robjects as robjects
r = robjects.r
import pandas.rpy.common as com
r.library("impute")
r_data = com.convert_to_r_matrix(data)
r_imputedData = r['impute.knn'](r_data)
npImputedData = np.asarray(r_imputedData[0])
imputedData = pd.DataFrame(npImputedData)
imputedData.index = data.index
imputedData.columns = data.columns
return imputedData
def drop_missing_genes(data, naLimit=0.5):
"""Filters a data frame to weed out cols with missing data"""
thresh = len(data.columns) * (1 - naLimit)
return data.dropna(thresh=thresh, axis="rows")
def drop_missing_samples(data, naLimit=0.8):
"""Filters a data frame to weed out cols with missing data"""
thresh = len(data.index) * (1 - naLimit)
return data.dropna(thresh=thresh, axis="columns")
def translate_negative_cols(data):
"""Translate the minimum value of each col to 1"""
data = data.replace([np.inf, -np.inf], np.nan) #replace infinities
return data + np.abs(np.min(data)) + 1
def clean_data(data):
"""convenience function to trannslate the data before analysis"""
if not data.empty:
# data = log_data(translate_negative_cols(data))
data = drop_missing_samples(data)
return data