def test_GetTypeStrings(self):
headers = ['pk', 'a', 'b', 'c']
colTypes = [(int, 2), (int, 3), (float, 5), (str, 10)]
self.assertEqual(
DbUtils.GetTypeStrings(headers, colTypes),
['pk integer', 'a integer', 'b double precision', 'c varchar(10)'])
self.assertEqual(
DbUtils.GetTypeStrings(headers, colTypes, keyCol='pk'),
['pk integer not null primary key', 'a integer', 'b double precision', 'c varchar(10)'])
def FingerprintsFromDetails(details, reportFreq=10):
data = None
if details.dbName and details.tableName:
from rdkit.Dbase.DbConnection import DbConnect
from rdkit.Dbase import DbInfo
from rdkit.ML.Data import DataUtils
try:
conn = DbConnect(details.dbName, details.tableName)
except Exception:
import traceback
error('Problems establishing connection to database: %s|%s\n' %
(details.dbName, details.tableName))
traceback.print_exc()
if not details.idName:
details.idName = DbInfo.GetColumnNames(details.dbName,
details.tableName)[0]
dataSet = DataUtils.DBToData(details.dbName,
details.tableName,
what='%s,%s' %
(details.idName, details.smilesName))
idCol = 0
smiCol = 1
elif details.inFileName and details.useSmiles:
from rdkit.ML.Data import DataUtils
conn = None
if not details.idName:
details.idName = 'ID'
try:
dataSet = DataUtils.TextFileToData(
details.inFileName,
onlyCols=[details.idName, details.smilesName])
except IOError:
import traceback
error('Problems reading from file %s\n' % (details.inFileName))
traceback.print_exc()
idCol = 0
smiCol = 1
elif details.inFileName and details.useSD:
conn = None
dataset = None
if not details.idName:
details.idName = 'ID'
dataSet = []
try:
s = Chem.SDMolSupplier(details.inFileName)
except Exception:
import traceback
error('Problems reading from file %s\n' % (details.inFileName))
traceback.print_exc()
else:
while 1:
try:
m = s.next()
except StopIteration:
break
if m:
dataSet.append(m)
if reportFreq > 0 and not len(dataSet) % reportFreq:
message('Read %d molecules\n' % (len(dataSet)))
if details.maxMols > 0 and len(
dataSet) >= details.maxMols:
break
for i, mol in enumerate(dataSet):
if mol.HasProp(details.idName):
nm = mol.GetProp(details.idName)
else:
nm = mol.GetProp('_Name')
dataSet[i] = (nm, mol)
else:
dataSet = None
fps = None
if dataSet and not details.useSD:
data = dataSet.GetNamedData()
if not details.molPklName:
fps = FingerprintsFromSmiles(data, idCol, smiCol,
**details.__dict__)
else:
fps = FingerprintsFromPickles(data, idCol, smiCol,
**details.__dict__)
elif dataSet and details.useSD:
fps = FingerprintsFromMols(dataSet, **details.__dict__)
if fps:
if details.outFileName:
outF = open(details.outFileName, 'wb+')
for i in range(len(fps)):
pickle.dump(fps[i], outF)
outF.close()
dbName = details.outDbName or details.dbName
if details.outTableName and dbName:
from rdkit.Dbase.DbConnection import DbConnect
from rdkit.Dbase import DbUtils, DbModule
conn = DbConnect(dbName)
#
# We don't have a db open already, so we'll need to figure out
# the types of our columns...
#
colTypes = DbUtils.TypeFinder(data, len(data), len(data[0]))
typeStrs = DbUtils.GetTypeStrings(
[details.idName, details.smilesName],
colTypes,
keyCol=details.idName)
cols = '%s, %s %s' % (typeStrs[0], details.fpColName,
DbModule.binaryTypeName)
# FIX: we should really check to see if the table
# is already there and, if so, add the appropriate
# column.
#
# create the new table
#
if details.replaceTable or \
details.outTableName.upper() not in [x.upper() for x in conn.GetTableNames()]:
conn.AddTable(details.outTableName, cols)
#
# And add the data
#
for ID, fp in fps:
tpl = ID, DbModule.binaryHolder(fp.ToBinary())
conn.InsertData(details.outTableName, tpl)
conn.Commit()
return fps
