Re: [Rdkit-discuss] ConstrainedEmbed issue

[Pavel Polishchuk]

Thank you, Sereina.









I provided the code and output below.

Code:

data = [('1.mol', 'C[C@@H]1CCCCC1=O', 'C[C@@H]1CC[C@H](O)CC1=O'),


for i, (mol_fname, smi_template, smi_child) in enumerate(data):

    print('iteration', i)

    mode = 'read template mol file, no AddHs'
    print(mode)
    mol_template = Chem.MolFromMolFile(mol_fname)
    mol_child = Chem.MolFromSmiles(smi_child)
    try:
        mol = AllChem.ConstrainedEmbed(mol_child, mol_template)
        print(mol.GetProp('EmbedRMS'))
    except ValueError as e:
        print(e)

    mode = 'read template mol file, AddHs'
    print(mode)
    mol_template = Chem.MolFromMolFile(mol_fname)
    mol_child = Chem.MolFromSmiles(smi_child)
    try:
        mol = AllChem.ConstrainedEmbed(Chem.AddHs(mol_child), mol_template)
        print(mol.GetProp('EmbedRMS'))
    except ValueError as e:
        print(e)

    mode = 'embed template mol in rdkit, no AddHs'
    print(mode)
    mol_template = Chem.MolFromSmiles(smi_template)
    AllChem.EmbedMolecule(mol_template)
    mol_child = Chem.MolFromSmiles(smi_child)
    try:
        mol = AllChem.ConstrainedEmbed(mol_child, mol_template)
        print(mol.GetProp('EmbedRMS'))
    except ValueError as e:
        print(e)

Output:

iteration 0
read template mol file, no AddHs
Could not embed molecule.
read template mol file, AddHs
0.05014807519735495
embed template mol in rdkit, no AddHs
0.12358989886023371

iteration 1
read template mol file, no AddHs
0.057937898735270194
read template mol file, AddHs

embed template mol in rdkit, no AddHs
0.1012757033705761


Pavel.



On 07/07/2020 21:41, Sunhwan Jo wrote:

I understand importance of addition of hydrogens to get a reasonable 3D coordinates. But the situation may be not that simple. 1. Addition of hydrogen is only required for custom coordinates supplied from an external file. If coordinates of a template is generated with rdkit embedding it works without addition of explicit hydrogens. 2. I found an opposite example where addition of hydrogens breaks constrained embedding if custom coordinates of a template is used. And again if I generate coordinates of a template by rdkit everything is OK without addition of Hs. These suggest that there is some issue with custom coordinates usage for constrained embedding.         ('2.mol', 'CCCCCCCC[C@@H](CCC)NC(=O)c1ccc(F)cc1', 'CCCC[C@H](CCC[C@@H](CCC)NC(=O)c1ccc(F)cc1)NC(=O)c1ccco1')] Could not embed molecule.     # <-- here rdkit spends a lot of time but fails

Makes sense :)

On Jul 7, 2020, at 12:35 PM, Sereina Riniker <sereina.rini...@gmail.com <mailto:sereina.rini...@gmail.com>> wrote:

Dear Pavel and Sunhwan,

Please note that hydrogens should always be added for the embedding algorithm to work properly (i.e. it’s not a walk around but what should be done). See also Section “Working with 3D Molecules” in https://www.rdkit.org/docs/GettingStartedInPython.html

Best regards,
Sereina

On 7 Jul 2020, at 21:26, Sunhwan Jo <sunhw...@gmail.com <mailto:sunhw...@gmail.com>> wrote:

The reason constraint embed didn’t work is the molecule simply can’t be embedded using the rdkit’s algorithm.

In [25]: mol_child = Chem.MolFromSmiles('C[C@@H]1CC[C@H](O)CC1=O')

In [26]: AllChem.EmbedMolecule(mol_child)
Out[26]: -1

See more discussion here:
https://github.com/rdkit/rdkit/issues/2996



someone with more in-depth insight can help here.


Anyway, for a walk around, adding H seems to do the trick:

The SMILES you posted looks valid to me and doesn’t look that complicated, but the anyway I think somehow the RDKit’s algorithm tripped up and couldn’t finish embedding without some help. Hope

In [39]: mol = AllChem.AddHs(mol_child)

In [40]: AllChem.EmbedMolecule(mol)
Out[40]: 0 # worked

In [41]: AllChem.ConstrainedEmbed(mol, mol_parent)
Out[41]: <rdkit.Chem.rdchem.Mol at 0x7fe8000f6f80> # also worked

Sunhwan

On Jul 7, 2020, at 12:36 AM, Pavel Polishchuk <pavel_polishc...@ukr.net <mailto:pavel_polishc...@ukr.net>> wrote:

Hi all,


C[C@@H]1CC[C@H](O)CC1=O.



mol_parent = Chem.MolFromMolFile('1.mol')
mol_child = Chem.MolFromSmiles('C[C@@H]1CC[C@H](O)CC1=O')
try:
    mol = AllChem.ConstrainedEmbed(mol_child, mol_parent)
except ValueError as e:
    print(e)

  If I add explicit hydrogens the issue disappears.

mol_parent = Chem.MolFromMolFile('1.mol')
mol_child = Chem.MolFromSmiles('C[C@@H]1CC[C@H](O)CC1=O')
mol = AllChem.ConstrainedEmbed(Chem.AddHs(mol_child), mol_parent)

  If I do not use pre-defined coordinates - everything works well.

mol_parent = Chem.MolFromSmiles('C[C@@H]1CCCCC1=O')
AllChem.EmbedMolecule(mol_parent)
mol_child = Chem.MolFromSmiles('C[C@@H]1CC[C@H](O)CC1=O')
mol = AllChem.ConstrainedEmbed(mol_child, mol_parent)



Kind regards,
Pavel.
<1.mol>_______________________________________________
Rdkit-discuss mailing list

  I have an issue with ConstrainedEmbed and I cannot figure out what exactly causes this.   I have a molecule C[C@@H]1CCCCC1=O with 3D coordinates in 1.mol file (attached). And I want to generate coordinates for another structure with this core -   This is usual way which causes issue with embedding and the corresponding error.   Does ugly coordinates in 1.mol file cause the embedding issue? Or the issue is caused by some implicit properties of a molecule? How to solve this properly? Rdkit-discuss@lists.sourceforge.net <mailto:Rdkit-discuss@lists.sourceforge.net>

https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

_______________________________________________
Rdkit-discuss mailing list

Rdkit-discuss@lists.sourceforge.net <mailto:Rdkit-discuss@lists.sourceforge.net>

https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

1.mol
Description: MOL mdl chemical test

2.mol
Description: MOL mdl chemical test

_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss