Re: [Rdkit-discuss] implicit conversion of smart pointer with version 2018_03_4
Greg,
I figure it out. Adding const should work.
const Atom *atom = (*pCurrentROMol)[*atBegin];
Thanks.
Yingfeng
On Wed, Dec 12, 2018 at 10:21 AM Yingfeng Wang wrote:
> Greg,
>
> I figure out the problem. Removing "->get()" can solve the problem.
>
> Here is the updated version.
>
> boost::tie(firstB,lastB) = pCurrentROMol->getEdges();
>
> while(firstB!=lastB)
>
> {
>
> auto bond = (*pCurrentROMol)[*firstB];
>
> ...
>
> sEdgeLabel += ((queryIsBondInRing(bond)) ? "ring|" : "linear|" );
>
> ...
>
> ++firstB;
>
> }
>
>
> The original version was
>
>
> boost::tie(firstB,lastB) = pCurrentROMol->getEdges();
>
> while(firstB!=lastB)
>
> {
>
> boost::shared_ptr bond = (*pCurrentROMol)[*firstB];
>
> ...
>
> sEdgeLabel += ((queryIsBondInRing(bond.get())) ? "ring|" : "linear|"
> );
>
> ...
>
> ++firstB;
>
> }
>
>
> However, I still have a question. Why "Atom *atom" and "Bond *bond" do not
> work? Yes, "auto" solves the problem, but I feel it just covers the
> essential issue.
>
>
> Thanks.
>
> Yingfeng
>
>
>
>
> On Wed, Dec 12, 2018 at 9:50 AM Yingfeng Wang wrote:
>
>> Greg,
>>
>> Thanks. The first way does not work. I got the following error message.
>>
>> *Database.cpp:148:15: **error: **cannot initialize a variable of type
>> 'RDKit::Atom *'*
>>
>> * with an rvalue of type 'const RDKit::Atom *'*
>>
>> Atom *atom = (*pCurrentROMol)[*atBegin];
>>
>>
>> The second way, which uses "auto", works. I also updated the code for
>> bond and got a new error with the following error message.
>>
>> *Database.cpp:164:49: **error: **no member named 'get' in 'RDKit::Bond'*
>>
>> sEdgeLabel += ((queryIsBondInRing(bond->get())) ? "ring|" :
>> "linear|" );
>>
>> * ^*
>>
>> 1 error generated.
>>
>> Could you please confirm that the updated class Bond removes method
>> get()? If this is the case, which method should I use?
>>
>> Again, thank you very much for your help!
>>
>> Yingfeng
>>
>>
>>
>> On Wed, Dec 12, 2018 at 12:58 AM Greg Landrum
>> wrote:
>>
>>> Hi Yingfeng,
>>>
>>> As part of the move over to Modern C++ we also changed the way atoms and
>>> bonds are stored in molecules: you now get raw pointers back instead of
>>> smart pointers.
>>> If you change your code from:
>>> boost::shared_ptr atom = (*pCurrentROMol)[*atBegin];
>>> to:
>>> Atom *atom = (*pCurrentROMol)[*atBegin];
>>> or, even simpler:
>>> auto atom = (*pCurrentROMol)[*atBegin];
>>>
>>> things should work.
>>> -greg
>>>
>>>
>>> On Wed, Dec 12, 2018 at 12:36 AM Yingfeng Wang
>>> wrote:
>>>
I am using the C++ library of RDKit on Mac. My C++ code works with
RDKit_2017_09_3. However, after I switch to RDKit 2018_03_4, I got the
following error when compiling my C++ source code.
*Database.cpp:148:33: **error: **no viable conversion from 'const
RDKit::Atom *' to*
* 'boost::shared_ptr'*
boost::shared_ptr atom = (*pCurrentROMol)[*atBegin];
*^ ~~*
*/usr/local/Cellar/boost/1.68.0/include/boost/smart_ptr/shared_ptr.hpp:358:21:
**note: *
candidate constructor not viable: no known conversion from
'const RDKit::Atom *' to 'boost::detail::sp_nullptr_t' (aka
'nullptr_t')
for 1st argument
BOOST_CONSTEXPR shared_ptr( boost::detail::sp_nullptr_t )
BOOST_SP_N...
*^*
*/usr/local/Cellar/boost/1.68.0/include/boost/smart_ptr/shared_ptr.hpp:422:5:
**note: *
candidate constructor not viable: no known conversion from
'const RDKit::Atom *' to 'const boost::shared_ptr
&' for 1st
argument
shared_ptr( shared_ptr const & r ) BOOST_SP_NOEXCEPT : px( r.px ),
p...
*^*
I am using Clang on Mac. The version information is given as follows.
g++ -v
Configured with: --prefix=/Library/Developer/CommandLineTools/usr
--with-gxx-include-dir=/Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/usr/include/c++/4.2.1
Apple LLVM version 10.0.0 (clang-1000.10.44.4)
Target: x86_64-apple-darwin18.2.0
Thread model: posix
I notice that "Starting with the 2018_03 release, the RDKit core C++
code is written in modern C++; for this release that means C++11. "
Actually, I also use -std=c++11 when compiling my C++ source code. I
also tested RDKit 2018_09_1 and got the similar error. I am wondering how
to fix this problem.
Thanks.
Yingfeng
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>>>
___
Rdkit-discuss mailing list
Re: [Rdkit-discuss] RGroup matching in RGroup decomposition code
HI Paolo
That's cool thanks. This will also maybe help me in trying to solve my problem
of R-group label numbering not taking into account the actual R-group numbering
(ie if a molecule has R8 and R5 as sole R-group definitions then they get R1,R2
labels).
I also was in contact with Brian Kelley and he suggested to fix it in the
underlying codebase so I hope this will be fixed in the next version :)
Cheers
Nik
From: Paolo Tosco
Sent: Thursday, December 13, 2018 11:09 AM
To: Stiefl, Nikolaus ; RDKit Discuss
Subject: Re: [Rdkit-discuss] RGroup matching in RGroup decomposition code
Hi Nik,
There is a way to achieve what you describe, even though it is slightly
cumbersome:
from rdkit import Chem
from rdkit.Chem import rdmolops
from rdkit.Chem.Draw import MolsToGridImage, IPythonConsole
from rdkit.Chem.rdRGroupDecomposition import (
RGroupDecomposition, RGroupDecompositionParameters)
smis = ['Cc1ccnc(O)c1', 'Cc1cc(Cl)ccn1', 'Nc1n1',
'Nc1ccc(Br)cn1', 'c1ccncc1']
mols = [Chem.MolFromSmiles(smi) for smi in smis]
MolsToGridImage(mols)
[cid:image001.png@01D49312.8A22D520]
params = RGroupDecompositionParameters()
# rather than using the built-in flag we will manually
# adjust the query in two steps using AdjustQueryProperties()
params.onlyMatchAtRGroups = False
# just atom number the rgroups
core1 = Chem.MolFromSmiles('n1ccc([*:2])cc([*:1])1')
# make dummies queries
core1_params = rdmolops.AdjustQueryParameters()
core1_params.makeDummiesQueries = True
core1_params.adjustDegree = False
core1 = rdmolops.AdjustQueryProperties(core1, core1_params)
# change the atoms connected to the dummies into dummies
former_atomic_nums = {}
for b in core1.GetBonds():
if (b.GetBeginAtom().GetAtomicNum() == 0):
a = b.GetEndAtom()
elif (b.GetEndAtom().GetAtomicNum() == 0):
a = b.GetBeginAtom()
else:
continue
former_atomic_nums[a.GetIdx()] = a.GetAtomicNum()
a.SetAtomicNum(0)
# this has the same effect as setting onlyMatchAtRGroups to True
# but we can avoid applying it the atoms connected to the R groups
core1_params.adjustHeavyDegreeFlags = Chem.ADJUST_IGNOREDUMMIES
core1_params.makeDummiesQueries = False
core1_params.adjustDegree = False
core1_params.adjustHeavyDegree = True
core1 = rdmolops.AdjustQueryProperties(core1, core1_params)
# restore the original atomic numbers
for i, an in former_atomic_nums.items():
core1.GetAtomWithIdx(i).SetAtomicNum(an)
rg1 = RGroupDecomposition(core1, params)
failMols = []
for m in mols:
res = rg1.Add(m)
if res < 0:
failMols.append(m)
rg1.Process()
True
print("FailedMols: %s"%" ".join([Chem.MolToSmiles(m) for m in failMols]))
FailedMols: Nc1ccc(Br)cn1
core1
[cid:image002.png@01D49312.8A22D520]
d = rg1.GetRGroupsAsColumns(asSmiles=False)
MolsToGridImage(d['Core'])
[cid:image003.png@01D49312.8A22D520]
MolsToGridImage(d['R1'])
[cid:image004.png@01D49312.8A22D520]
MolsToGridImage(d['R2'])
[cid:image005.png@01D49312.8A22D520]
Hope that helps, cheers
p.
On 12/11/18 11:01, Stiefl, Nikolaus wrote:
Hi all,
I was playing around with the RGroup decomposition code and must say that I am
pretty impressed by it. The fact that one can directly work with a MDL R-group
file and that the output is a pandasDataFrame makes analysis really slick -
well done !
However, one thing that irritates me is the fact that seemingly when I have
R-groups defined in my core and enforce matching only at R-groups then
molecules having hydrogen atoms in that position are ignored in the "Add" step.
I would expect those to be included as long as the molecules don't have
additional heavy atoms in positions that are not defined as R-groups in the
core.
__ snip
from rdkit import Chem
from rdkit.Chem.rdRGroupDecomposition import RGroupDecomposition,
RGroupDecompositionParameters
smis = ['Cc1ccnc(O)c1', 'Cc1cc(Cl)ccn1', 'Nc1n1', 'Nc1ccc(Br)cn1',
'c1ccncc1']
mols = [Chem.MolFromSmiles(smi) for smi in smis]
params = RGroupDecompositionParameters()
params.onlyMatchAtRGroups = True
# just atom number the rgroups
core1 = Chem.MolFromSmiles('n1ccc([*:2])cc([*:1])1')
rg1 = RGroupDecomposition(core1, params)
failMols = []
for m in mols:
res = rg1.Add(m)
if res < 0:
failMols.append(m)
rg1.Process()
print("FailedMols: %s"%" ".join([Chem.MolToSmiles(m) for m in failMols]))
end snip
the output shows that molecules 3-5 are not included at the "Add" step
>> FailedMols: Nc1n1 Nc1ccc(Br)cn1 c1ccncc1
For molecules 4 (the 5-bromo substituted aminopyridine) I agree, however I
don't understand how I can make sure mols 3 and 5 are also included ... is
there a magic parameter that I can set?
Cheers
Nik
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
Re: [Rdkit-discuss] assigning bond orders from reference: problems with COOH group
Hi Mariana, if you modify your Jupyter notebook as in https://gist.github.com/ptosco/d807c64df9f277e92284bc5a7cecbc1d then it should work also in the presence of graph isomorphisms. Cheers, p. On 12/13/18 12:18, Mariana Assmann wrote: Hi all, I have docking poses (from AutoDockVina) in a format that does not provide bond information, and there is no information on hydrogens atoms unless they are important for hydrogens bonds. I was assigning the bond orders from the original reference structure or it's smiles string using the methods in https://gist.github.com/ptosco/4844d3635cf14d11e5e14381993915c1 using AllChem.AssignBondOrdersFromTemplate() and adding the original hydrogens back on. This approach does not work well when I have -COOH groups present. In AssignBondOrdersFromTemplate, the substructure match almost always assigns the oxygens the opposite way so that my resulting structure is -C(OH)=OH which can't be sanitized in the end. I would like to keep the configuration of the docking output. Is there a way to correct that in an automated way? I have quite a few poses with that problem and can't do that manually. I have uploaded a gist with an example problem here: https://gist.github.com/c4asma/b338321238d1924cc2964612ed278e9e Here, oxygen with atom number 46 has the hydrogen, and number 45 should be assigned a double bond (input 7). But, after AllChem.AssignBondOrdersFromTemplate() oxygen 46 is assigned the double bond (input 15) Thank you, Mariana ___ 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] Question about (generic) Murcko frameworks
Hi there, is it normal to get, for the same set of smiles, more generic Murcko frameworks that normal frameworks? When running this: ### Generate framework for a SMILES, handling for errors def framecheck(s): try: return Chem.MolToSmiles(ms.GetScaffoldForMol(Chem.MolFromSmiles(s))) except: pass ### Generate generic framework for a SMILES, handling for errors def gframecheck(s): try: return Chem.MolToSmiles(ms.MakeScaffoldGeneric(Chem.MolFromSmiles(s))) except: pass # Count unique frameworks fraq = [framecheck(s) for s in smis] fraq = list(set(fraq)) len(fraq) # Count unique generic frameworks gfraq = [gframecheck(s) for s in smis] gfraq = list(set(gfraq)) len(gfraq) I get for the attached set of smiles 1431 frameworks and 2207 generic frameworks. The generic frameworks are supposed to have all atom types set to C and all bonds to single, so I would expect less generic frameworks. Is it necessary some sort of canonicalization? Thanks in advance Gonzalo examp.smi Description: Binary data ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
[Rdkit-discuss] assigning bond orders from reference: problems with COOH group
Hi all, I have docking poses (from AutoDockVina) in a format that does not provide bond information, and there is no information on hydrogens atoms unless they are important for hydrogens bonds. I was assigning the bond orders from the original reference structure or it's smiles string using the methods in https://gist.github.com/ptosco/4844d3635cf14d11e5e14381993915c1 using AllChem.AssignBondOrdersFromTemplate() and adding the original hydrogens back on. This approach does not work well when I have -COOH groups present. In AssignBondOrdersFromTemplate, the substructure match almost always assigns the oxygens the opposite way so that my resulting structure is -C(OH)=OH which can't be sanitized in the end. I would like to keep the configuration of the docking output. Is there a way to correct that in an automated way? I have quite a few poses with that problem and can't do that manually. I have uploaded a gist with an example problem here: https://gist.github.com/c4asma/b338321238d1924cc2964612ed278e9e Here, oxygen with atom number 46 has the hydrogen, and number 45 should be assigned a double bond (input 7). But, after AllChem.AssignBondOrdersFromTemplate() oxygen 46 is assigned the double bond (input 15) Thank you, Mariana ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] RGroup matching in RGroup decomposition code
Hi Nik,
There is a way to achieve what you describe, even though it is slightly
cumbersome:
from rdkit import Chem
from rdkit.Chem import rdmolops
from rdkit.Chem.Draw import MolsToGridImage, IPythonConsole
from rdkit.Chem.rdRGroupDecomposition import (
RGroupDecomposition, RGroupDecompositionParameters)
smis = ['Cc1ccnc(O)c1', 'Cc1cc(Cl)ccn1', 'Nc1n1',
'Nc1ccc(Br)cn1', 'c1ccncc1']
mols = [Chem.MolFromSmiles(smi) for smi in smis]
MolsToGridImage(mols)
params = RGroupDecompositionParameters()
# rather than using the built-in flag we will manually
# adjust the query in two steps using AdjustQueryProperties()
params.onlyMatchAtRGroups = False
# just atom number the rgroups
core1 = Chem.MolFromSmiles('n1ccc([*:2])cc([*:1])1')
# make dummies queries
core1_params = rdmolops.AdjustQueryParameters()
core1_params.makeDummiesQueries = True
core1_params.adjustDegree = False
core1 = rdmolops.AdjustQueryProperties(core1, core1_params)
# change the atoms connected to the dummies into dummies
former_atomic_nums = {}
for b in core1.GetBonds():
if (b.GetBeginAtom().GetAtomicNum() == 0):
a = b.GetEndAtom()
elif (b.GetEndAtom().GetAtomicNum() == 0):
a = b.GetBeginAtom()
else:
continue
former_atomic_nums[a.GetIdx()] = a.GetAtomicNum()
a.SetAtomicNum(0)
# this has the same effect as setting onlyMatchAtRGroups to True
# but we can avoid applying it the atoms connected to the R groups
core1_params.adjustHeavyDegreeFlags = Chem.ADJUST_IGNOREDUMMIES
core1_params.makeDummiesQueries = False
core1_params.adjustDegree = False
core1_params.adjustHeavyDegree = True
core1 = rdmolops.AdjustQueryProperties(core1, core1_params)
# restore the original atomic numbers
for i, an in former_atomic_nums.items():
core1.GetAtomWithIdx(i).SetAtomicNum(an)
rg1 = RGroupDecomposition(core1, params)
failMols = []
for m in mols:
res = rg1.Add(m)
if res < 0:
failMols.append(m)
rg1.Process()
True
print("FailedMols: %s"%" ".join([Chem.MolToSmiles(m) for m in failMols]))
FailedMols: Nc1ccc(Br)cn1
core1
d = rg1.GetRGroupsAsColumns(asSmiles=False)
MolsToGridImage(d['Core'])
MolsToGridImage(d['R1'])
MolsToGridImage(d['R2'])
Hope that helps, cheers
p.
On 12/11/18 11:01, Stiefl, Nikolaus wrote:
Hi all,
I was playing around with the RGroup decomposition code and must say
that I am pretty impressed by it. The fact that one can directly work
with a MDL R-group file and that the output is a pandasDataFrame makes
analysis really slick – well done !
However, one thing that irritates me is the fact that seemingly when I
have R-groups defined in my core and enforce matching only at R-groups
then molecules having hydrogen atoms in that position are ignored in
the “Add” step. I would expect those to be included as long as the
molecules don’t have additional heavy atoms in positions that are not
defined as R-groups in the core.
__ snip
from rdkit import Chem
from rdkit.Chem.rdRGroupDecomposition import RGroupDecomposition,
RGroupDecompositionParameters
smis = ['Cc1ccnc(O)c1', 'Cc1cc(Cl)ccn1', 'Nc1n1', 'Nc1ccc(Br)cn1',
'c1ccncc1']
mols = [Chem.MolFromSmiles(smi) for smi in smis]
params = RGroupDecompositionParameters()
params.onlyMatchAtRGroups = True
# just atom number the rgroups
core1 = Chem.MolFromSmiles('n1ccc([*:2])cc([*:1])1')
rg1 = RGroupDecomposition(core1, params)
failMols = []
for m in mols:
res = rg1.Add(m)
if res < 0:
failMols.append(m)
rg1.Process()
print("FailedMols: %s"%" ".join([Chem.MolToSmiles(m) for m in failMols]))
end snip
the output shows that molecules 3-5 are not included at the “Add” step
>> FailedMols: Nc1n1 Nc1ccc(Br)cn1 c1ccncc1
For molecules 4 (the 5-bromo substituted aminopyridine) I agree,
however I don’t understand how I can make sure mols 3 and 5 are also
included … is there a magic parameter that I can set?
Cheers
Nik
___
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
