import numpy as np
import pubchempy
import rdkit.Chem.Lipinski as Lipinksy
from IPython.display import Image, display
from rdkit import Chem, RDConfig
from rdkit.Chem import AllChem, Draw
from rdkit.Chem.Draw import IPythonConsole

Ибупрофен

ibu=Chem.MolFromSmiles("CC(C)CC1=CC=C(C=C1)C(C)C(=O)O")
AllChem.Compute2DCoords(ibu)
display(ibu)

Параметры для правила Лепински

print(Lipinksy.NumHDonors(ibu))
print(Lipinksy.NumHAcceptors(ibu))
print(Lipinksy.rdMolDescriptors.CalcExactMolWt(ibu))
print(Lipinksy.rdMolDescriptors.CalcCrippenDescriptors(ibu))

1
1
206.130679816
(3.073200000000001, 61.03480000000004)

def is_lepinsky_valid(compound):
    """Проверяет валидность структуры по правилам Lepinsky"""
    try:
        return (
            (Lipinksy.NumHDonors(compound) <= 5)
            and (Lipinksy.NumHAcceptors(compound) <= 10)
            and (Lipinksy.rdMolDescriptors.CalcExactMolWt(ibu) < 500)
            and (Lipinksy.rdMolDescriptors.CalcCrippenDescriptors(ibu)[0] <= 5)
        )
    except:
        return False

Примеры модификации ибупрофена для клик-реакции и результат самой реакции

ibu_modified = Chem.MolFromSmiles("C(C(C)CC1=CC=C(C=C1)C(C)C(=O)O)#C")
AllChem.Compute2DCoords(ibu_modified)
display(ibu_modified)

ibu_clicked = Chem.MolFromSmiles("CC(C)CC1=CC=C(C=C1)C(C(C(N=N1)=CN1))C(=O)O")
AllChem.Compute2DCoords(ibu_clicked)
display(ibu_clicked)

ibu_modified = Chem.MolFromSmiles("C(C(C)CC1=CC=C(C=C1)C(C)C(=O)O)#C")
AllChem.Compute2DCoords(ibu_modified)
display(ibu_modified)

ibu_clicked = Chem.MolFromSmiles("C1(=C[NH]N=N1)C(C)CC2=CC=C(C=C2)C(C)C(=O)O")
AllChem.Compute2DCoords(ibu_clicked)
display(ibu_clicked)
# Дальше будем вставлять все в эту структуру. Ее шаблон вставки N1N=N-C=C1C(C)CC2=CC=C(C=C2)C(C)C(=O)O

ibu_modified = Chem.MolFromSmiles("CC(C)CC1=C(C=C(C=C1)C(C)C(=O)O)C#C")
AllChem.Compute2DCoords(ibu_modified)
display(ibu_modified)

ibu_clicked = Chem.MolFromSmiles("CC(C)CC1=C(C=C(C=C1)C(C)C(=O)O)C2=CN=N[NH]2")
AllChem.Compute2DCoords(ibu_clicked)
display(ibu_clicked)

Загрузка азидов

compounds = []
per_page = 10**5
for smiles in ["N=N=N", "NN#N",]:
    for i in range(200):
        try:
            a = pubchempy.get_properties(
                properties="CanonicalSMILES", 
                identifier=smiles, 
                namespace="smiles", 
                searchtype="substructure",
                RingsNotEmbedded=True,
                listkey_count=per_page,
                listkey_start=i*per_page
            )
        except:
            break
        print("Retrieved page {} of {} search".format(i+1, smiles))
        compounds.extend(a)

Retrieved page 1 of N=N=N search
Retrieved page 2 of N=N=N search
Retrieved page 3 of N=N=N search
Retrieved page 4 of N=N=N search
Retrieved page 5 of N=N=N search
Retrieved page 6 of N=N=N search
Retrieved page 7 of N=N=N search
Retrieved page 8 of N=N=N search
Retrieved page 9 of N=N=N search
Retrieved page 10 of N=N=N search
Retrieved page 11 of N=N=N search
Retrieved page 12 of N=N=N search
Retrieved page 13 of N=N=N search
Retrieved page 14 of N=N=N search
Retrieved page 15 of N=N=N search
Retrieved page 16 of N=N=N search
Retrieved page 1 of NN#N search

# Всего соединений
len(compounds)

1601375

Фильтрация подходящих азидов

Возьмем небольшие азиды с группой N=[N+]=[N-]

filtered_smiles = [
    x["CanonicalSMILES"]
    for x in compounds
    if ("N=[N+]=[N-]" in x["CanonicalSMILES"]) and len(x["CanonicalSMILES"]) <= 30
]
len(filtered_smiles)

251084

Пример азида с ибупрофеном

ibu_clicked = Chem.MolFromSmiles(
    filtered_smiles[0].replace("N=[N+]=[N-]", "N1N=N-C=C1C(C)CC2=CC=C(C=C2)C(C)C(=O)O")
)
AllChem.Compute2DCoords(ibu_clicked)
display(ibu_clicked)

Преобразование и получение валидных структур, соответствующих правилам

valid_structures = []
for fsmi in filtered_smiles:
    struct = Chem.MolFromSmiles(
        fsmi.replace("N=[N+]=[N-]", "N1N=N-C=C1C(C)CC2=CC=C(C=C2)C(C)C(=O)O")
    )
    if is_lepinsky_valid(struct):
        valid_structures.append(struct)

RDKit ERROR: [15:48:14] Explicit valence for atom # 8 Cl, 3, is greater than permitted
RDKit ERROR: [15:48:14] Explicit valence for atom # 1 Br, 3, is greater than permitted

наиболее частая причина невалидности структур -- наличие ионов хлора на конце smiles

len(valid_structures)

250852

Draw.MolsToGridImage(
    valid_structures[:30], useSVG=True, molsPerRow=3, subImgSize=(250, 120)
)

Similarity maps

from rdkit.Chem.Draw import SimilarityMaps

fp = SimilarityMaps.GetMorganFingerprint(valid_structures[5], fpType="bv")

fig, maxweight = SimilarityMaps.GetSimilarityMapForFingerprint(
    ibu, valid_structures[5], SimilarityMaps.GetMorganFingerprint
)

3D view

Chem.MolToSmiles(valid_structures[5])

'CC(Cc1ccc(C(C)C(=O)O)cc1)c1cnnn1CCC1CCN(C)C(=O)O1'

struct = Chem.MolFromSmiles("CC(Cc1ccc(C(C)C(=O)O)cc1)c1cnnn1CCC1CCN(C)C(=O)O1")

m3d = Chem.AddHs(struct)
Chem.AllChem.EmbedMolecule(m3d)
AllChem.MMFFOptimizeMolecule(m3d, maxIters=500, nonBondedThresh=200)

0

import nglview as nv

nv.show_rdkit(m3d)