taxol = taxol[0]
type(taxol) ### pubchempy.Compound
taxol = Chem.MolFromSmiles(taxol.canonical_smiles)
type(taxol) ### rdkit.Chem.rdchem.Mol
### Morganフィンガープリント
bitI_morgan = {}
fp_morgan = AllChem.GetMorganFingerprintAsBitVect(taxol, 2, bitInfo=bitI_morgan)
### RDKitフィンガープリント
bitI_rdkit = {}
fp_rdkit = Chem.RDKFingerprint(taxol, bitInfo=bitI_rdkit)
print(fp_morgan.GetNumBits(),fp_morgan.GetNumOnBits()) ### 2048 86
print(len(bitI_morgan)) ### 86
print(len(fp_rdkit), len(bitI_rdkit.keys())) ### (2048, 1444)
for key in list(bitI_morgan.keys())[:5]:
print(bitI_morgan[key])
### Morgan可視化
morgan_turples = ((taxol, bit, bitI_morgan) for bit in list(bitI_morgan.keys())[:12])
Draw.DrawMorganBits(morgan_turples, molsPerRow=4, legends=['bit: '+str(x) for x in list(bitI_morgan.keys())[:12]])
### RDKit可視化
rdkit_turples = ((taxol, bit, bitI_rdkit) for bit in list(bitI_rdkit.keys())[:12])
Draw.DrawRDKitBits(rdkit_turples, molsPerRow=4, legends=['bit: '+str(x) for x in list(bitI_rdkit.keys())[:12]])
# 3. 灰色：说明该原子是脂肪烃原子
# # 3.8 可视化拓扑指纹中的bit
# 拓扑指纹也称为RDKit指纹，其调用函数Chem.RDKFingerprint(mol)
mol = Chem.MolFromSmiles('c1cccnc1C')
rdkbi = {}
rdkfp = Chem.RDKFingerprint(mol, maxPath=2, bitInfo=rdkbi)
print(list(rdkbi.keys()))
# [5, 161, 294, 330, 633, 684, 744, 808, 842, 930, 1026, 1027, 1060, 1649, 1909]
# 可视化展示
rdkbi_tuple = [(mol, bit, rdkbi) for bit in list(rdkbi.keys())]
img = Draw.DrawRDKitBits(
rdkbi_tuple,
molsPerRow=4,
subImgSize=(200, 200),
legends=list(
map(str, list(rdkbi.keys()))
)
)
with open('/drug_development/studyRdkit/st_rdcit/img/mol27.svg', 'w+') as outf:
outf.write(img)
# # 3.9 基于分子指纹挑选差异较大的分子
# 药物虚拟筛选中关键步骤挑选分子，比如筛选获得前1000个分子， 由于成本、时间等因素你想挑选100个分子进行活性测试， 如果你直接挑选前100个分子进行测试，命中率可能会降低。 一般流程是对1000个分子进行聚类，然后每一类里面挑选一个分子（或者中心分子）， 这样可以提高分子骨架的多样性，从而提供虚拟筛选的成功率。
ms = [x for x in Chem.SDMolSupplier(
'/Users/zeoy/st/drug_development/st_rdcit/2d.sdf')]
while ms.count(None):
ms.remove(None)
fps = [AllChem.GetMorganFingerprint(x, 3) for x in ms]
