Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Paolo,
argh... I thought if you are setting params.sanitize=False then you don't
want sanitization.
Apparently, you need it for the mols but skipping only aromatization.
I guess, I slowly start to understand...
What you explained here and in your github example I tried to find something
similar in the RDKit documentation or in the web... without too much success.
Wouldn't this be an essential step in substructure search or even a FAQ?
Well, now I took a larger list and I got as many hits as I expected. Happy End!
Thank you very much for your kind help!
Theo.
Am 20.05.2020 um 15:22 schrieb Paolo Tosco:
> Hi Theo,
>
> that's because you omitted the sanitization step completely, so the molecule
> is missing crucial information for the SubstructureMatch to do a proper job.
>
> If you put back sanitization, only leaving out the aromatization step, things
> work as expected.
> Also, you do not need to create pattern again from SMILES, you can make a
> copy of the molecule that you have already created and sanitized using the
> Chem.Mol copy constructor.
>
> from rdkit import Chem
>
> smiles_strings = '''
> N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
> C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
> '''
>
> smiles_list = smiles_strings.splitlines()[1:]
> print(smiles_list)
>
> params = Chem.SmilesParserParams()
> params.sanitize=False
>
> mols = [Chem.MolFromSmiles(x,params) for x in smiles_list]
> for m in mols:
> Chem.SanitizeMol(m, Chem.SANITIZE_ALL ^ Chem.SANITIZE_SETAROMATICITY)
>
> pattern = Chem.Mol(mols[0])
>
> query_params = Chem.AdjustQueryParameters()
> query_params.makeBondsGeneric = True
> query_params.aromatizeIfPossible = False
> query_params.adjustDegree = False
> query_params.adjustHeavyDegree = False
> pattern_generic_bonds = Chem.AdjustQueryProperties(pattern,query_params)
>
> matches = [idx for idx,m in enumerate(mols) if
> m.HasSubstructMatch(pattern_generic_bonds)]
> print("{} of {}: {}".format(len(matches),len(smiles_list),matches))
>
> $ python3 SubstructMatch2.py
>
> ['N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3', 'C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3']
> 2 of 2: [0, 1]
>
> Cheers,
> p.
>
> On 20/05/2020 09:50, theozh wrote:
>> from rdkit import Chem
>>
>> smiles_strings = '''
>> N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
>> C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
>> '''
>>
>> smiles_list = smiles_strings.splitlines()[1:]
>> print(smiles_list)
>>
>> params = Chem.SmilesParserParams()
>> params.sanitize=False
>>
>> mols = [Chem.MolFromSmiles(x,params) for x in smiles_list]
>>
>> pattern = Chem.MolFromSmiles(smiles_list[0],params)
>>
>> query_params = Chem.AdjustQueryParameters()
>> query_params.makeBondsGeneric = True
>> query_params.aromatizeIfPossible = False
>> query_params.adjustDegree = False
>> query_params.adjustHeavyDegree = False
>> pattern_generic_bonds = Chem.AdjustQueryProperties(pattern,query_params)
>>
>> matches = [idx for idx,m in enumerate(mols) if
>> m.HasSubstructMatch(pattern_generic_bonds)]
>> print("{} of {}: {}".format(len(matches),len(smiles_list),matches))
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Theo,
that's because you omitted the sanitization step completely, so the
molecule is missing crucial information for the SubstructureMatch to do
a proper job.
If you put back sanitization, only leaving out the aromatization step,
things work as expected.
Also, you do not need to create pattern again from SMILES, you can make
a copy of the molecule that you have already created and sanitized using
the Chem.Mol copy constructor.
from rdkit import Chem
smiles_strings = '''
N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
'''
smiles_list = smiles_strings.splitlines()[1:]
print(smiles_list)
params = Chem.SmilesParserParams()
params.sanitize=False
mols = [Chem.MolFromSmiles(x,params) for x in smiles_list]
for m in mols:
Chem.SanitizeMol(m, Chem.SANITIZE_ALL ^ Chem.SANITIZE_SETAROMATICITY)
pattern = Chem.Mol(mols[0])
query_params = Chem.AdjustQueryParameters()
query_params.makeBondsGeneric = True
query_params.aromatizeIfPossible = False
query_params.adjustDegree = False
query_params.adjustHeavyDegree = False
pattern_generic_bonds = Chem.AdjustQueryProperties(pattern,query_params)
matches = [idx for idx,m in enumerate(mols) if
m.HasSubstructMatch(pattern_generic_bonds)]
print("{} of {}: {}".format(len(matches),len(smiles_list),matches))
$ python3 SubstructMatch2.py
['N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3', 'C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3']
2 of 2: [0, 1]
Cheers,
p.
On 20/05/2020 09:50, theozh wrote:
from rdkit import Chem
smiles_strings = '''
N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
'''
smiles_list = smiles_strings.splitlines()[1:]
print(smiles_list)
params = Chem.SmilesParserParams()
params.sanitize=False
mols = [Chem.MolFromSmiles(x,params) for x in smiles_list]
pattern = Chem.MolFromSmiles(smiles_list[0],params)
query_params = Chem.AdjustQueryParameters()
query_params.makeBondsGeneric = True
query_params.aromatizeIfPossible = False
query_params.adjustDegree = False
query_params.adjustHeavyDegree = False
pattern_generic_bonds = Chem.AdjustQueryProperties(pattern,query_params)
matches = [idx for idx,m in enumerate(mols) if
m.HasSubstructMatch(pattern_generic_bonds)]
print("{} of {}: {}".format(len(matches),len(smiles_list),matches))
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Paolo,
sorry, I made a typo (makeBondGeneric instead of makeBondsGeneric) that's why
the bonds weren't UNSPECIFIED.
The following examples seem to work fine now for these two SMILES, the first
structure will be found in the second one.
C12=CC=CN1NCCC2
and
C12C=CC=C(C=C3)C=1N3NCC2
However, there is another example where it still doesn't work with this code.
See my code below.
The two SMILES
N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
and
C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
actually describe the identical structure, but were drawn in a different way in
ChemDraw. As a consequence the SMILES are different which shouldn't be a
problem. But if I put these SMILES into the code below the first one won't
match the second one and the other way around as well.
I must be doing something horribly wrong.
Do I have to canonicalize the SMILES first?
Isn't there a good tutorial on substructure search with RDKit and all its
options and frequently asked questions and tons of examples?
best,
Theo.
### start of code
from rdkit import Chem
smiles_strings = '''
N12N3C(CC4=CC=CC(NC=C2)=C14)=CC=C3
C12=CC=CC3=C1N(N4C=CC=C4C2)C=CN3
'''
smiles_list = smiles_strings.splitlines()[1:]
print(smiles_list)
params = Chem.SmilesParserParams()
params.sanitize=False
mols = [Chem.MolFromSmiles(x,params) for x in smiles_list]
pattern = Chem.MolFromSmiles(smiles_list[0],params)
query_params = Chem.AdjustQueryParameters()
query_params.makeBondsGeneric = True
query_params.aromatizeIfPossible = False
query_params.adjustDegree = False
query_params.adjustHeavyDegree = False
pattern_generic_bonds = Chem.AdjustQueryProperties(pattern,query_params)
matches = [idx for idx,m in enumerate(mols) if
m.HasSubstructMatch(pattern_generic_bonds)]
print("{} of {}: {}".format(len(matches),len(smiles_list),matches))
### end of code
Am 19.05.2020 um 18:30 schrieb Paolo Tosco:
> Hi Theo,
>
> I don't think the RDKit version should make a difference; did you notice that
> rdmolops.AdjustQueryProperties() does not modify the molecule in place, but
> rather returns a modified copy?
>
> pattern_generic_bonds = Chem.AdjustQueryProperties(pattern, query_params)
>
> That might be the reason. Also, only pattern_generic_bonds will have
> UNSPECIFIED bonds, the mols will still have SINGLE and DOUBLE bonds.
>
> Feel free to contact me off-list if you need help with the above.
>
> Cheers,
> p.
>
> On 19/05/2020 17:01, theozh wrote:
>> Hi Paolo,
>>
>> thank you very much for your detailed answer.
>> I tried to reproduce your last suggestion (but I don't have Jupyter
>> Notebook).
>> However, my bonds are still SINGLE and DOUBLE instead of UNSPECIFIED.
>> Does this maybe depend on the RDKit Version, I have 2019.03... ?
>>
>> Maybe, I should update and need to investigate further.
>> Theo.
>>
>>
>> Am 19.05.2020 um 16:44 schrieb Paolo Tosco:
>>> Hi Theo,
>>>
>>> the lack of match is due to different aromaticity flags on atoms and bonds
>>> in the larger molecule.
>>>
>>> This gist provides some explanation and a possible solution:
>>>
>>> https://gist.github.com/ptosco/e410e45278b94e8f047ff224193d7788
>>>
>>> Cheers,
>>> p.
>>>
>>> On 19/05/2020 14:13, theozh wrote:
Dear RDKit-users,
I would like to do a very simple substructure search.
The chapter 3.5 "Substructure Searching" in RDKit Documentation
(2019.09.1) is pretty short and doesn't point to a solution. So far, I've
learned that you can create your search pattern via Chem.MolFromSmiles()
or Chem.MolFromSmarts().
In the below copy minimal example, I want to use the first SMILES in
the list as search pattern. I expect 2 matches but I get either 1 or 0
matches. So, I'm doing something wrong. What am I missing?
Is it about implicit/explicit aromatic and aliphatic bonds or some
explicit/implicit hydrogen?
How to find the first structure in both SMILES?
thank you for any hints,
Theo.
### simple substructure search (but doesn't find what is expected)
from rdkit import Chem
smiles_strings = '''
C12=CC=CN1NCCC2
C12=CC=CC(C=C3)=C1N3NCC2
'''
smiles_list = smiles_strings.splitlines()[1:]
print(smiles_list)
pattern = Chem.MolFromSmiles(smiles_list[0]) # MolFromSmiles
matches = [x for x in smiles_list if
Chem.MolFromSmiles(x).HasSubstructMatch(pattern)]
print(len(matches)) # result: 1, why not 2?
pattern = Chem.MolFromSmarts(smiles_list[0]) # MolFromSmarts
matches = [x for x in smiles_list if
Chem.MolFromSmiles(x).HasSubstructMatch(pattern)]
print(len(matches)) # result: 0, why not 2?
### end of code
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
___
Rdkit-discuss mailing list
Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Theo, I don't think the RDKit version should make a difference; did you notice that rdmolops.AdjustQueryProperties() does not modify the molecule in place, but rather returns a modified copy? pattern_generic_bonds = Chem.AdjustQueryProperties(pattern, query_params) That might be the reason. Also, only pattern_generic_bonds will have UNSPECIFIED bonds, the mols will still have SINGLE and DOUBLE bonds. Feel free to contact me off-list if you need help with the above. Cheers, p. On 19/05/2020 17:01, theozh wrote: Hi Paolo, thank you very much for your detailed answer. I tried to reproduce your last suggestion (but I don't have Jupyter Notebook). However, my bonds are still SINGLE and DOUBLE instead of UNSPECIFIED. Does this maybe depend on the RDKit Version, I have 2019.03... ? Maybe, I should update and need to investigate further. Theo. Am 19.05.2020 um 16:44 schrieb Paolo Tosco: Hi Theo, the lack of match is due to different aromaticity flags on atoms and bonds in the larger molecule. This gist provides some explanation and a possible solution: https://gist.github.com/ptosco/e410e45278b94e8f047ff224193d7788 Cheers, p. On 19/05/2020 14:13, theozh wrote: Dear RDKit-users, I would like to do a very simple substructure search. The chapter 3.5 "Substructure Searching" in RDKit Documentation (2019.09.1) is pretty short and doesn't point to a solution. So far, I've learned that you can create your search pattern via Chem.MolFromSmiles() or Chem.MolFromSmarts(). In the below copy minimal example, I want to use the first SMILES in the list as search pattern. I expect 2 matches but I get either 1 or 0 matches. So, I'm doing something wrong. What am I missing? Is it about implicit/explicit aromatic and aliphatic bonds or some explicit/implicit hydrogen? How to find the first structure in both SMILES? thank you for any hints, Theo. ### simple substructure search (but doesn't find what is expected) from rdkit import Chem smiles_strings = ''' C12=CC=CN1NCCC2 C12=CC=CC(C=C3)=C1N3NCC2 ''' smiles_list = smiles_strings.splitlines()[1:] print(smiles_list) pattern = Chem.MolFromSmiles(smiles_list[0]) # MolFromSmiles matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 1, why not 2? pattern = Chem.MolFromSmarts(smiles_list[0]) # MolFromSmarts matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 0, why not 2? ### end of code ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Paolo, thank you very much for your detailed answer. I tried to reproduce your last suggestion (but I don't have Jupyter Notebook). However, my bonds are still SINGLE and DOUBLE instead of UNSPECIFIED. Does this maybe depend on the RDKit Version, I have 2019.03... ? Maybe, I should update and need to investigate further. Theo. Am 19.05.2020 um 16:44 schrieb Paolo Tosco: > Hi Theo, > > the lack of match is due to different aromaticity flags on atoms and bonds in > the larger molecule. > > This gist provides some explanation and a possible solution: > > https://gist.github.com/ptosco/e410e45278b94e8f047ff224193d7788 > > Cheers, > p. > > On 19/05/2020 14:13, theozh wrote: >> Dear RDKit-users, >> >> I would like to do a very simple substructure search. >> The chapter 3.5 "Substructure Searching" in RDKit Documentation (2019.09.1) >> is pretty short and doesn't point to a solution. So far, I've learned that >> you can create your search pattern via Chem.MolFromSmiles() or >> Chem.MolFromSmarts(). >> >> In the below copy minimal example, I want to use the first SMILES in >> the list as search pattern. I expect 2 matches but I get either 1 or 0 >> matches. So, I'm doing something wrong. What am I missing? >> Is it about implicit/explicit aromatic and aliphatic bonds or some >> explicit/implicit hydrogen? >> How to find the first structure in both SMILES? >> >> thank you for any hints, >> Theo. >> >> ### simple substructure search (but doesn't find what is expected) >> from rdkit import Chem >> >> smiles_strings = ''' >> C12=CC=CN1NCCC2 >> C12=CC=CC(C=C3)=C1N3NCC2 >> ''' >> smiles_list = smiles_strings.splitlines()[1:] >> print(smiles_list) >> >> pattern = Chem.MolFromSmiles(smiles_list[0]) # MolFromSmiles >> matches = [x for x in smiles_list if >> Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] >> print(len(matches)) # result: 1, why not 2? >> >> pattern = Chem.MolFromSmarts(smiles_list[0]) # MolFromSmarts >> matches = [x for x in smiles_list if >> Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] >> print(len(matches)) # result: 0, why not 2? >> ### end of code >> >> >> ___ >> Rdkit-discuss mailing list >> Rdkit-discuss@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Hi Theo, the lack of match is due to different aromaticity flags on atoms and bonds in the larger molecule. This gist provides some explanation and a possible solution: https://gist.github.com/ptosco/e410e45278b94e8f047ff224193d7788 Cheers, p. On 19/05/2020 14:13, theozh wrote: Dear RDKit-users, I would like to do a very simple substructure search. The chapter 3.5 "Substructure Searching" in RDKit Documentation (2019.09.1) is pretty short and doesn't point to a solution. So far, I've learned that you can create your search pattern via Chem.MolFromSmiles() or Chem.MolFromSmarts(). In the below copy minimal example, I want to use the first SMILES in the list as search pattern. I expect 2 matches but I get either 1 or 0 matches. So, I'm doing something wrong. What am I missing? Is it about implicit/explicit aromatic and aliphatic bonds or some explicit/implicit hydrogen? How to find the first structure in both SMILES? thank you for any hints, Theo. ### simple substructure search (but doesn't find what is expected) from rdkit import Chem smiles_strings = ''' C12=CC=CN1NCCC2 C12=CC=CC(C=C3)=C1N3NCC2 ''' smiles_list = smiles_strings.splitlines()[1:] print(smiles_list) pattern = Chem.MolFromSmiles(smiles_list[0]) # MolFromSmiles matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 1, why not 2? pattern = Chem.MolFromSmarts(smiles_list[0]) # MolFromSmarts matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 0, why not 2? ### end of code ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
[Rdkit-discuss] Substructure search issue with aliphatic/aromatic bonds
Dear RDKit-users, I would like to do a very simple substructure search. The chapter 3.5 "Substructure Searching" in RDKit Documentation (2019.09.1) is pretty short and doesn't point to a solution. So far, I've learned that you can create your search pattern via Chem.MolFromSmiles() or Chem.MolFromSmarts(). In the below copy minimal example, I want to use the first SMILES in the list as search pattern. I expect 2 matches but I get either 1 or 0 matches. So, I'm doing something wrong. What am I missing? Is it about implicit/explicit aromatic and aliphatic bonds or some explicit/implicit hydrogen? How to find the first structure in both SMILES? thank you for any hints, Theo. ### simple substructure search (but doesn't find what is expected) from rdkit import Chem smiles_strings = ''' C12=CC=CN1NCCC2 C12=CC=CC(C=C3)=C1N3NCC2 ''' smiles_list = smiles_strings.splitlines()[1:] print(smiles_list) pattern = Chem.MolFromSmiles(smiles_list[0]) # MolFromSmiles matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 1, why not 2? pattern = Chem.MolFromSmarts(smiles_list[0]) # MolFromSmarts matches = [x for x in smiles_list if Chem.MolFromSmiles(x).HasSubstructMatch(pattern)] print(len(matches)) # result: 0, why not 2? ### end of code ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
