def test7ScoreMolecules(self):
suppl = Chem.SmilesMolSupplierFromText('\n'.join(self.smiList2), ',',
0, -1, 0)
cat = BuildFragmentCatalog.BuildCatalog(suppl,
minPath=1,
reportFreq=20)
assert cat.GetNumEntries() == 21
assert cat.GetFPLength() == 21
scores, obls = BuildFragmentCatalog.ScoreMolecules(suppl,
cat,
acts=self.list2Acts,
reportFreq=20)
for i in range(len(self.list2Obls)):
assert tuple(
obls[i]) == self.list2Obls[i], '%d: %s != %s' % (i, str(
obls[i]), str(self.list2Obls[i]))
scores2 = BuildFragmentCatalog.ScoreFromLists(obls,
suppl,
cat,
acts=self.list2Acts,
reportFreq=20)
for i in range(len(scores)):
assert (scores[i] == scores2[i]).all(), '%d: %s != %s' % (
i, str(scores[i]), str(scores2[i]))
def test6Builder(self):
suppl = Chem.SmilesMolSupplierFromText('\n'.join(self.smiList2),
',',0,-1,0)
cat = BuildFragmentCatalog.BuildCatalog(suppl,minPath=1,reportFreq=20)
assert cat.GetNumEntries()==21
assert cat.GetFPLength()==21
self._testBits(cat)
def test9Issue116(self):
smiList = ['Cc1ccccc1']
suppl = Chem.SmilesMolSupplierFromText('\n'.join(smiList), ',', 0, -1, 0)
cat = BuildFragmentCatalog.BuildCatalog(suppl, minPath=2, maxPath=2)
assert cat.GetFPLength() == 2
assert cat.GetBitDescription(0) == 'ccC'
fpgen = FragmentCatalog.FragFPGenerator()
mol = Chem.MolFromSmiles('Cc1ccccc1')
fp = fpgen.GetFPForMol(mol, cat)
assert fp[0]
assert fp[1]
mol = Chem.MolFromSmiles('c1ccccc1-c1ccccc1')
fp = fpgen.GetFPForMol(mol, cat)
assert not fp[0]
assert fp[1]
def test8MolRanks(self):
suppl = Chem.SmilesMolSupplierFromText('\n'.join(self.smiList2), ',',
0, -1, 0)
cat = BuildFragmentCatalog.BuildCatalog(suppl,
minPath=1,
reportFreq=20)
assert cat.GetNumEntries() == 21
assert cat.GetFPLength() == 21
# new InfoGain ranking:
bitInfo, fps = BuildFragmentCatalog.CalcGains(suppl,
cat,
topN=10,
acts=self.list2Acts,
reportFreq=20,
biasList=(1, ))
entry = bitInfo[0]
assert int(entry[0]) == 0
assert cat.GetBitDescription(int(entry[0])) == 'C<-O>C'
assert feq(entry[1], 0.4669)
entry = bitInfo[1]
assert int(entry[0]) in (2, 6)
txt = cat.GetBitDescription(int(entry[0]))
self.assertTrue(txt in ('C<-O>CC', 'C<-O>=C'), txt)
assert feq(entry[1], 0.1611)
entry = bitInfo[6]
assert int(entry[0]) == 16
assert cat.GetBitDescription(int(entry[0])) == 'C=CC<-O>'
assert feq(entry[1], 0.0560)
# standard InfoGain ranking:
bitInfo, fps = BuildFragmentCatalog.CalcGains(suppl,
cat,
topN=10,
acts=self.list2Acts,
reportFreq=20)
entry = bitInfo[0]
assert int(entry[0]) == 0
assert cat.GetBitDescription(int(entry[0])) == 'C<-O>C'
assert feq(entry[1], 0.4669)
entry = bitInfo[1]
assert int(entry[0]) == 5
assert cat.GetBitDescription(int(entry[0])) == 'C=CC'
assert feq(entry[1], 0.2057)