aboutsummaryrefslogtreecommitdiff
path: root/scripts/maintenance
diff options
context:
space:
mode:
authorBonfaceKilz2020-08-26 22:23:40 +0300
committerBonfaceKilz2020-08-26 22:23:40 +0300
commit768de71f09b116cad0f42c792c22a7dc410ea2f4 (patch)
tree6cf95e463e7b07b0a8cb3104385e54b81bef99ba /scripts/maintenance
parent24bed3fc4f3a69c24074e9c5829a46b66702cbbf (diff)
downloadgenenetwork2-768de71f09b116cad0f42c792c22a7dc410ea2f4.tar.gz
Replace tabs with spaces
* scripts/maintenance/readProbeSetMean_v7.py: Run python-autopep8 on this file.
Diffstat (limited to 'scripts/maintenance')
-rwxr-xr-xscripts/maintenance/readProbeSetMean_v7.py250
1 files changed, 125 insertions, 125 deletions
diff --git a/scripts/maintenance/readProbeSetMean_v7.py b/scripts/maintenance/readProbeSetMean_v7.py
index 59a51cf9..a540796a 100755
--- a/scripts/maintenance/readProbeSetMean_v7.py
+++ b/scripts/maintenance/readProbeSetMean_v7.py
@@ -9,19 +9,17 @@ import sys
import MySQLdb
import getpass
import time
-#import pdb
-#pdb.set_trace()
########################################################################
def translateAlias(str):
- if str == "B6":
- return "C57BL/6J"
- elif str == "D2":
- return "DBA/2J"
- else:
- return str
+ if str == "B6":
+ return "C57BL/6J"
+ elif str == "D2":
+ return "DBA/2J"
+ else:
+ return str
########################################################################
#
@@ -29,23 +27,25 @@ def translateAlias(str):
#
########################################################################
+
dataStart = 1
-GeneChipId = int( input("Enter GeneChipId:") )
-ProbeSetFreezeId = int( input("Enter ProbeSetFreezeId:") )
+GeneChipId = int(input("Enter GeneChipId:"))
+ProbeSetFreezeId = int(input("Enter ProbeSetFreezeId:"))
input_file_name = input("Enter file name with suffix:")
fp = open("%s" % input_file_name, 'rb')
try:
- passwd = getpass.getpass('Please enter mysql password here : ')
- con = MySQLdb.Connect(db='db_webqtl', host='localhost', user='username', passwd=passwd)
+ passwd = getpass.getpass('Please enter mysql password here : ')
+ con = MySQLdb.Connect(db='db_webqtl', host='localhost',
+ user='username', passwd=passwd)
- db = con.cursor()
- print("You have successfully connected to mysql.\n")
+ db = con.cursor()
+ print("You have successfully connected to mysql.\n")
except:
- print("You entered incorrect password.\n")
- sys.exit(0)
+ print("You entered incorrect password.\n")
+ sys.exit(0)
time0 = time.time()
@@ -65,25 +65,25 @@ header = list(map(string.strip, header))
nfield = len(header)
line = fp.readline()
-kj=0
+kj = 0
while line:
- line2 = string.split(string.strip(line), '\t')
- line2 = list(map(string.strip, line2))
- if len(line2) != nfield:
- print(("Error : " + line))
- isCont = 0
+ line2 = string.split(string.strip(line), '\t')
+ line2 = list(map(string.strip, line2))
+ if len(line2) != nfield:
+ print(("Error : " + line))
+ isCont = 0
- GeneList.append(line2[0])
- line = fp.readline()
+ GeneList.append(line2[0])
+ line = fp.readline()
- kj+=1
- if kj%100000 == 0:
- print(('checked ', kj, ' lines'))
+ kj += 1
+ if kj % 100000 == 0:
+ print(('checked ', kj, ' lines'))
GeneList = sorted(map(string.lower, GeneList))
-
-if isCont==0:
- sys.exit(0)
+
+if isCont == 0:
+ sys.exit(0)
print(('used ', time.time()-time0, ' seconds'))
@@ -104,15 +104,15 @@ header = list(map(translateAlias, header))
header = header[dataStart:]
Ids = []
for item in header:
- try:
- db.execute('select Id from Strain where Name = "%s"' % item)
- Ids.append(db.fetchall()[0][0])
- except:
- print((item, 'does not exist, check the if the strain name is correct'))
- isCont=0
+ try:
+ db.execute('select Id from Strain where Name = "%s"' % item)
+ Ids.append(db.fetchall()[0][0])
+ except:
+ print((item, 'does not exist, check the if the strain name is correct'))
+ isCont = 0
-if isCont==0:
- sys.exit(0)
+if isCont == 0:
+ sys.exit(0)
print(('used ', time.time()-time0, ' seconds'))
@@ -130,59 +130,60 @@ line2 = string.split(string.strip(line), '\t')
line2 = list(map(string.strip, line2))
PId = line2[0]
-db.execute('select Id from ProbeSet where Name="%s" and ChipId=%d' % (PId, GeneChipId) )
+db.execute('select Id from ProbeSet where Name="%s" and ChipId=%d' %
+ (PId, GeneChipId))
results = db.fetchall()
IdStr = 'TargetId'
-if len(results)>0:
- IdStr = 'Name'
+if len(results) > 0:
+ IdStr = 'Name'
##---- get Name/TargetId list from database ----##
-db.execute('select distinct(%s) from ProbeSet where ChipId=%d order by %s' % (IdStr, GeneChipId, IdStr))
+db.execute('select distinct(%s) from ProbeSet where ChipId=%d order by %s' % (
+ IdStr, GeneChipId, IdStr))
results = db.fetchall()
-
+
Names = []
for item in results:
- Names.append(item[0])
-
+ Names.append(item[0])
+
print(Names)
Names = sorted(map(string.lower, Names))
-
##---- compare genelist with names ----##
-x=y=0
-x1=-1
-GeneList2=[]
-while x<len(GeneList) and y<len(Names):
- if GeneList[x]==Names[y]:
- x += 1
- y += 1
- elif GeneList[x]<Names[y]:
- if x!=x1:
- GeneList2.append(GeneList[x])
- x1 = x
- x += 1
- elif GeneList[x]>Names[y]:
- y += 1
-
- if x%100000==0:
- print(('check Name, checked %d lines'%x))
-
-while x<len(GeneList):
- GeneList2.append(GeneList[x])
- x += 1
-
-isCont=1
+x = y = 0
+x1 = -1
+GeneList2 = []
+while x < len(GeneList) and y < len(Names):
+ if GeneList[x] == Names[y]:
+ x += 1
+ y += 1
+ elif GeneList[x] < Names[y]:
+ if x != x1:
+ GeneList2.append(GeneList[x])
+ x1 = x
+ x += 1
+ elif GeneList[x] > Names[y]:
+ y += 1
+
+ if x % 100000 == 0:
+ print(('check Name, checked %d lines' % x))
+
+while x < len(GeneList):
+ GeneList2.append(GeneList[x])
+ x += 1
+
+isCont = 1
ferror = open("ProbeSetError.txt", "wb")
for item in GeneList2:
- ferror.write(item + " doesn't exist \n")
- print((item, " doesn't exist, check if the ProbeSet name is correct"))
- isCont = 0
-
-if isCont==0:
- sys.exit(0)
+ ferror.write(item + " doesn't exist \n")
+ print((item, " doesn't exist, check if the ProbeSet name is correct"))
+ isCont = 0
+
+if isCont == 0:
+ sys.exit(0)
print(('used ', time.time()-time0, ' seconds'))
@@ -195,11 +196,12 @@ print('getting ProbeSet/Id')
#---- get Name/Id map ----#
-db.execute('select %s, Id from ProbeSet where ChipId=%d order by %s' % (IdStr, GeneChipId, IdStr))
+db.execute('select %s, Id from ProbeSet where ChipId=%d order by %s' %
+ (IdStr, GeneChipId, IdStr))
results = db.fetchall()
NameIds = {}
for item in results:
- NameIds[item[0]] = item[1]
+ NameIds[item[0]] = item[1]
print(('used ', time.time()-time0, ' seconds'))
@@ -220,53 +222,51 @@ kj = 0
values1 = []
values2 = []
while line:
- line2 = string.split(string.strip(line), '\t')
- line2 = list(map(string.strip, line2))
- PId = line2[0]
- recordId = NameIds[PId]
-
- maxDataId += 1
- datasorig = line2[dataStart:]
-
- ###### Data Table items ######
- i=0
- for item in datasorig:
- try:
- values1.append('(%d,%d,%s)' % (maxDataId, Ids[i], float(item)))
- except:
- pass
- i += 1
-
- values2.append("(%d,%d,%d)" % (ProbeSetFreezeId, recordId, maxDataId))
-
-
- ##---- insert into table ----##
- kj += 1
- if kj % 100 == 0:
- cmd = ','.join(values1)
- cmd = 'insert into ProbeSetData values %s' % cmd
- db.execute(cmd)
-
- cmd = ','.join(values2)
- cmd = 'insert into ProbeSetXRef(ProbeSetFreezeId, ProbeSetId, DataId) values %s' % cmd
- db.execute(cmd)
-
- values1=[]
- values2=[]
- print(('Inserted ', kj, ' lines'))
- print(('used ', time.time()-time0, ' seconds'))
-
- line = fp.readline()
-
-
-
-if len(values1)>0:
- cmd = ','.join(values1)
- cmd = 'insert into ProbeSetData values %s' % cmd
- db.execute(cmd)
-
- cmd = ','.join(values2)
- cmd = 'insert into ProbeSetXRef(ProbeSetFreezeId, ProbeSetId, DataId) values %s' % cmd
- db.execute(cmd)
+ line2 = string.split(string.strip(line), '\t')
+ line2 = list(map(string.strip, line2))
+ PId = line2[0]
+ recordId = NameIds[PId]
+
+ maxDataId += 1
+ datasorig = line2[dataStart:]
+
+ ###### Data Table items ######
+ i = 0
+ for item in datasorig:
+ try:
+ values1.append('(%d,%d,%s)' % (maxDataId, Ids[i], float(item)))
+ except:
+ pass
+ i += 1
+
+ values2.append("(%d,%d,%d)" % (ProbeSetFreezeId, recordId, maxDataId))
+
+ ##---- insert into table ----##
+ kj += 1
+ if kj % 100 == 0:
+ cmd = ','.join(values1)
+ cmd = 'insert into ProbeSetData values %s' % cmd
+ db.execute(cmd)
+
+ cmd = ','.join(values2)
+ cmd = 'insert into ProbeSetXRef(ProbeSetFreezeId, ProbeSetId, DataId) values %s' % cmd
+ db.execute(cmd)
+
+ values1 = []
+ values2 = []
+ print(('Inserted ', kj, ' lines'))
+ print(('used ', time.time()-time0, ' seconds'))
+
+ line = fp.readline()
+
+
+if len(values1) > 0:
+ cmd = ','.join(values1)
+ cmd = 'insert into ProbeSetData values %s' % cmd
+ db.execute(cmd)
+
+ cmd = ','.join(values2)
+ cmd = 'insert into ProbeSetXRef(ProbeSetFreezeId, ProbeSetId, DataId) values %s' % cmd
+ db.execute(cmd)
con.close()