def extract_Pharm2D(column,
minPointCount=2,
maxPointCount=3,
bins=[(0, 2), (2, 5), (5, 8)],
from_smiles=True):
"""Extract Pharm2D fingerprint
:param column: Pandas Series, containing smiles or RDKit mol object
:param minPointCount: int
:param maxPointCount: int
:param bins: lits of tuples
:param from_smiles: bool, indicate whether column contains smiles string
:return: feature_Pharm2D: Pandas Series, containing Pharm2D features
"""
sigFactory = SigFactory(featFactory,
minPointCount=minPointCount,
maxPointCount=maxPointCount,
trianglePruneBins=False)
sigFactory.SetBins(bins)
sigFactory.Init()
def get_Pharm2D(x):
mol = Chem.MolFromSmiles(x)
if (mol is None) or (len(mol.GetAtoms()) == 0):
return [0] * sigFactory.GetSigSize()
else:
return Generate.Gen2DFingerprint(mol, sigFactory)
fp = column.apply(lambda x: get_Pharm2D(x))
return np.array(list(fp))
def pharmacophore(mol, target):
i = 0
print('mol/target', mol, target)
mol.standardize()
target.standardize()
mol = str(mol)
mol = mol.replace('N(=O)O', '[N+](=O)[O-]')
mol = mol.replace('N(O)=O', '[N+]([O-])=O')
mol = mol.replace('n(O)', '[n+]([O-])')
target = str(target)
target = target.replace('N(=O)O', '[N+](=O)[O-]')
target = target.replace('N(O)=O', '[N+]([O-])=O')
target = target.replace('n(O)', '[n+]([O-])')
featfactory = load_factory()
sigfactory = SigFactory(featfactory,
minPointCount=2,
maxPointCount=3,
trianglePruneBins=False)
sigfactory.SetBins([(0, 2), (2, 5), (5, 8)])
sigfactory.Init()
mol1 = Chem.MolFromSmiles(mol)
mol2 = Chem.MolFromSmiles(target)
if mol1 and mol2:
fp1 = Generate.Gen2DFingerprint(mol1, sigfactory)
fp2 = Generate.Gen2DFingerprint(mol2, sigfactory)
sims = DataStructs.TanimotoSimilarity(fp1, fp2)
return sims
else:
i = i + 1
print('ошибка', i, mol)
return -100
def GetPharmacoPFPs(mol,
bins=[(i, i + 1) for i in range(20)],
minPointCount=2,
maxPointCount=2,
return_bitInfo=False):
'''
Note: maxPointCont with 3 is slowly
bins = [(i,i+1) for i in range(20)],
maxPonitCount=2 for large-scale computation
'''
MysigFactory = SigFactory(featFactory,
trianglePruneBins=False,
minPointCount=minPointCount,
maxPointCount=maxPointCount)
MysigFactory.SetBins(bins)
MysigFactory.Init()
res = Generate.Gen2DFingerprint(mol, MysigFactory)
arr = np.array(list(res)).astype(np.bool)
if return_bitInfo:
description = []
for i in range(len(res)):
description.append(MysigFactory.GetBitDescription(i))
return arr, description
return arr
def get_2Dfp(self, rdmols):
#: ファーマコフォアの初期設定
fdefName = r'ensemble/BaseFeatures.fdef'
featFactory = ChemicalFeatures.BuildFeatureFactory(fdefName)
sigFactory = SigFactory(featFactory, minPointCount=2, maxPointCount=3)
#: ファーマコフォア間の距離を離散化する
sigFactory.SetBins([(0, 2), (2, 4)])
sigFactory.Init()
fps1 = [
Generate.Gen2DFingerprint(mol, sigFactory).ToBitString()
for mol in rdmols
]
fps2 = [list(map(int, list(fps))) for fps in fps1]
fps3 = np.array(fps2)
return fps3
def pool_init(fdef_fname, bin_step):
global process_factory
global sig_factory
process_factory = ChemicalFeatures.BuildFeatureFactory(
fdef_fname) if fdef_fname else None
sig_factory = SigFactory(process_factory,
minPointCount=2,
maxPointCount=3,
trianglePruneBins=False)
q = []
i = bin_step
j = 0
while i < 20:
q.append((j, i))
j = i
i += bin_step
sig_factory.SetBins(q)
sig_factory.Init()
def CalculatePharm2D3pointFingerprint(mol, featFactory=featFactory):
"""
Calculate Pharm2D3point Fingerprints
"""
sigFactory_3point = SigFactory(featFactory,
minPointCount=3,
maxPointCount=3)
sigFactory_3point.SetBins([(0, 2), (2, 4), (4, 6), (6, 10)])
sigFactory_3point.Init()
res = Generate.Gen2DFingerprint(mol, sigFactory_3point)
res_keys = tuple(res.GetOnBits())
init_list = [0] * 2135
for res_key in res_keys:
init_list[res_key] = 1
BitVect = tuple(init_list)
return BitVect, res_keys, res
def read_file(fname, fcfp4, fdef_fname):
if not fcfp4:
featFactory = ChemicalFeatures.BuildFeatureFactory(fdef_fname)
sigFactory = SigFactory(featFactory, minPointCount=2, maxPointCount=3, trianglePruneBins=False)
sigFactory.SetBins([(0, 2), (2, 5), (5, 8)])
sigFactory.Init()
d = defaultdict(list)
with open(fname) as f:
for row in f:
smiles, ids, aff = row.strip().split('\t')
if smiles is not None:
mol = Chem.MolFromSmiles(smiles)
d['mol_name'].append(ids)
d['smiles'].append(smiles)
if fcfp4:
d['fingerprint'].append(AllChem.GetMorganFingerprint(mol, 2, useFeatures=True))
else:
d['fingerprint'].append(Generate.Gen2DFingerprint(mol, sigFactory))
return d
def CalculatePharm2D2pointFingerprint(mol, featFactory=featFactory):
"""
Calculate Pharm2D2point Fingerprints
"""
sigFactory_2point = SigFactory(featFactory,
minPointCount=2,
maxPointCount=2)
sigFactory_2point.SetBins([(0, 1), (1, 2), (2, 3), (3, 4), (4, 5), (5, 6),
(6, 7), (7, 8), (8, 9)])
sigFactory_2point.Init()
res = Generate.Gen2DFingerprint(mol, sigFactory_2point)
res_keys = tuple(res.GetOnBits())
init_list = [0] * 135
for res_key in res_keys:
init_list[res_key] = 1
BitVect = tuple(init_list)
return BitVect, res_keys, res
# 四、2D药效团指纹
# # 4.1 参数设置
# 对上面计算的性质进行组合可以用作分子的2D药效团。 药效团可以进一步转化为分子药效团指纹。
# 参考文件
fdefName = os.path.join(
RDConfig.RDDataDir,
'/drug_development/studyRdkit/st_rdcit/data/BaseFeatures.fdef')
# 实例化特征工厂
featFactory = ChemicalFeatures.BuildFeatureFactory(fdefName)
# 使用特征工厂再来构建指纹工厂signature，factory用于设置指纹参数
# 构建指纹工厂 ：
SigFactory(
featFactory, # 特征工厂
useCounts=False, # 默认False。False不考虑指纹频数，并生成SparseBitVect
minPointCount=2, # 默认为2.生成指纹时包括的最少的药效团数量。
maxPointCount=3, # 默认为3。生成指纹时包括的最多的药效团数量。
...)
sigFactory = SigFactory(featFactory, minPointCount=2, maxPointCount=3)
# 对拓扑距离进行分段
sigFactory.SetBins([(0, 2), (2, 5), (5, 8)])
# 每次修改参数后，都要初始化一下
sigFactory.Init()
# 计算指纹的长度
print('指纹长度=', sigFactory.GetSigSize()) # 指纹长度= 2988
# # 4.2 生成2D药效团指纹
# 指纹工厂中的参数设置完毕，接下来就可以生成2D指纹了。
# 计算2D药效团指纹 ：
Gen2DFingerprint(
mol, # 要计算指纹的mol对象
def get_factory():
featFactory = ChemicalFeatures.BuildFeatureFactoryFromString(fdef)
factory = SigFactory(featFactory, minPointCount=2, maxPointCount=3, useCounts=True, trianglePruneBins=False)
factory.SetBins(defaultBins)
factory.Init()
return factory
import os
import argparse
import pkg_resources
import pandas as pd
from multiprocessing import Pool
from rdkit.Chem import ChemicalFeatures
from rdkit.Chem.Pharm2D import Generate
from rdkit.Chem.Pharm2D.SigFactory import SigFactory
from .read_input import read_input
from .read_input import calc_max_tau
fdef_fname = pkg_resources.resource_filename(
__name__, 'pmapper_backlog/smarts_features.fdef')
featFactory = ChemicalFeatures.BuildFeatureFactory(fdef_fname)
sigFactory = SigFactory(featFactory,
minPointCount=2,
maxPointCount=3,
trianglePruneBins=False)
sigFactory.SetBins([(0, 2), (2, 5), (5, 8)])
sigFactory.Init()
def _ph_rdkit(mols_tup):
mol, name, act, _ = mols_tup
ph = Generate.Gen2DFingerprint(mol, sigFactory)
tmp = pd.DataFrame(columns=range(ph.GetNumBits()))
ph_bits = list(ph.GetOnBits())
for n_bit in ph_bits:
tmp.loc[name, n_bit] = 1
tmp.loc[name, 'mol_id'] = name
tmp.loc[name, 'act'] = act
tmp = tmp.fillna(0)
class AntidecoysSettings(Settings):
"""
A specialized class that holds the settings
for the anti-decoys algorithm.
"""
def __init__(
self,
source,
target,
storage_dir=os.path.abspath(
'antidecoys_data'
) # path to a directory where the results will be stored
,
max_threads=None # maximum number of threads to use in parallel computations
,
tree_params=None # custom parameters (same for both trees)
,
max_iters=100 # maximum number of iterations to spend looking for a single path
,
verbose=False # require verbose output
,
fg_bins=((0, 2), (2, 5), (5, 8)
) # distance bins in the pharmacophore fingerprint
,
fg_min_points=2 # min number of features encoded in the pharmacophore fingerprint
,
fg_max_points=3 # max number of features encoded in the pharmacophore fingerprint
,
min_accepted=1000 # minimum number of morphs the filter will accept on every iteration
,
common_bits_max_thrs=0.75 # maximum common bits percentage the filter will accept on every iteration
,
common_bits_mean_thrs=0.5 # if for the mols selected by the filter the mean common bits percentage falls below this value, antidecoys will be turned off
,
antidecoys_min_iters=10 # minimum number of iterations where antidecoys are optimized
,
antidecoys_max_iters=50 # maximum number of iterations where antidecoys are optimized
,
distance_thrs=0.2 # turn antidecoys filter off when the distance between two closest molecules from each tree gets below this value
):
super(AntidecoysSettings,
self).__init__(source, target, storage_dir, max_threads,
max_iters, tree_params, verbose)
self.fg_bins = fg_bins
"""
distance bins in the pharmacophore fingerprint (as described `here <http://www.rdkit.org/docs/GettingStartedInPython.html#d-pharmacophore-fingerprints>`_)
"""
self.fg_min_points = fg_min_points
"""
min number of features encoded in the pharmacophore fingerprint (as described `here <http://www.rdkit.org/docs/GettingStartedInPython.html#d-pharmacophore-fingerprints>`_)
"""
self.fg_max_points = fg_max_points
"""
max number of features encoded in the pharmacophore fingerprint (as described `here <http://www.rdkit.org/docs/GettingStartedInPython.html#d-pharmacophore-fingerprints>`_)
"""
self.min_accepted = min_accepted
"""
minimum number of morphs the antidecoys filter will accept on every iteration
"""
self.common_bits_max_thrs = common_bits_max_thrs
"""
maximum percentage of shared bits between a structure and the anti-fingerprint that the filter will accept on every iteration
"""
self.common_bits_mean_thrs = common_bits_mean_thrs
"""
if for the structures that survived the filter the mean common bits percentage falls below this value, antidecoys will be turned off
"""
self.antidecoys_min_iters = antidecoys_min_iters
"""
minimum number of iterations that will use the antidecoys filter
"""
self.antidecoys_max_iters = antidecoys_max_iters
"""
maximum number of iterations that will use the antidecoys filter
"""
self.distance_thrs = distance_thrs
"""
turn antidecoys filter off when the distance between two closest molecules from each tree gets below this value
"""
# stuff for the pharmacophore fingerprints
self._fdef_file = os.path.join(
RDConfig.RDDataDir,
'BaseFeatures.fdef') # get basic feature definitions
self._feature_factory = ChemicalFeatures.BuildFeatureFactory(
self._fdef_file) # make feature factory
self.signature_factory = SigFactory(
self._feature_factory,
minPointCount=self.fg_min_points,
maxPointCount=self.fg_max_points,
trianglePruneBins=False) # make signature factory
self.signature_factory.SetBins(self.fg_bins) # set the distance bins
self.signature_factory.Init()
def _init():
global labels, patts, factory
featFactory = ChemicalFeatures.BuildFeatureFactoryFromString(fdef)
factory = SigFactory(featFactory, minPointCount=2, maxPointCount=3)
factory.SetBins(defaultBins)
factory.Init()