Re: [Rdkit-discuss] UFF geometry optimization of lanthanide complexes
Hi Greg, Thanks for your reply, I'm aware that complexes of this kind are completely outside of this code's scope. I was just hoping it might still work here. Indeed, RDKit does not parse this molecule in SMILES format, but I was able to smuggle it through as MOL (in KNIME). But anyway, as only 6-valent Eu is defined in UFF, this is not the way forward for my purpose. Best wishes, Michal On 15 September 2015 at 10:54, Greg Landrum wrote: > Hi Michal, > > The problem here, I think, is that organometallic complexes like this one > involve bond types that are not well represented by SMILES, which really > assumes that a Lewis dot structure including shared electron pairs for all > bonds can be drawn. This is decidedly not the case here, where the molecule > can't even really properly be read in by the RDKit: > In [39]: m = > Chem.MolFromSmiles('[Eu]1234567OC(=CC(=[O]1)C)C.C(C=C(O2)C)(=[O]3)C.C(C=C(O4)C)(=[O]5)C.C1=[N]6C2=C(C=C1)C=CC1=C2[N]7=CC=C1') > [06:23:05] Explicit valence for atom # 5 O, 3, is greater than permitted > > The UFF parameters that are available for Eu are for Eu6+3: an > octahedrally coordinated Eu+3. Your complex has 8 connections to the Eu, so > it wouldn't be covered by the UFF parameters anyway. > > I just did a bit of playing around to see if I could construct a sample > molecule with a six-coordinate Eu+3 and make that work. I failed. This may > be an RDKit bug but I'm not quite sure. The corners of the code for dealing > with non-organic molecules are dark, dusty, and not particularly well > tested. > > Best, > -greg > > > > > On Mon, Sep 14, 2015 at 2:26 PM, Michal Krompiec < > michal.kromp...@gmail.com> wrote: > >> Hello, >> I was trying to generate 3D coordinates for an europium complex, >> [Eu(acac)3(phen)], with UFF, using RDKit nodes in KNIME (UFF is >> parametrized for lanthanides). Whereas the generation of coordinates seems >> to produce an almost sensible structure: >> [image: Inline images 3] >> >> subsequent geometry optimization does not: it moves the Eu atom way >> outside of the coordination sphere: >> [image: Inline images 4] >> >> Is it something with the bond types not specified correctly, or it is >> just not supposed to work with this type of molecules at all? The molecule >> is defined by the following SMILES (created with MarvinSketch): >> >> [Eu]1234567OC(=CC(=[O]1)C)C.C(C=C(O2)C)(=[O]3)C.C(C=C(O4)C)(=[O]5)C.C1=[N]6C2=C(C=C1)C=CC1=C2[N]7=CC=C1 >> The same result is obtained with La instead of Eu. >> Best wishes, >> Michal >> >> >> >> -- >> >> ___ >> 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] UFF geometry optimization of lanthanide complexes
Hi Michal, The problem here, I think, is that organometallic complexes like this one involve bond types that are not well represented by SMILES, which really assumes that a Lewis dot structure including shared electron pairs for all bonds can be drawn. This is decidedly not the case here, where the molecule can't even really properly be read in by the RDKit: In [39]: m = Chem.MolFromSmiles('[Eu]1234567OC(=CC(=[O]1)C)C.C(C=C(O2)C)(=[O]3)C.C(C=C(O4)C)(=[O]5)C.C1=[N]6C2=C(C=C1)C=CC1=C2[N]7=CC=C1') [06:23:05] Explicit valence for atom # 5 O, 3, is greater than permitted The UFF parameters that are available for Eu are for Eu6+3: an octahedrally coordinated Eu+3. Your complex has 8 connections to the Eu, so it wouldn't be covered by the UFF parameters anyway. I just did a bit of playing around to see if I could construct a sample molecule with a six-coordinate Eu+3 and make that work. I failed. This may be an RDKit bug but I'm not quite sure. The corners of the code for dealing with non-organic molecules are dark, dusty, and not particularly well tested. Best, -greg On Mon, Sep 14, 2015 at 2:26 PM, Michal Krompiec wrote: > Hello, > I was trying to generate 3D coordinates for an europium complex, > [Eu(acac)3(phen)], with UFF, using RDKit nodes in KNIME (UFF is > parametrized for lanthanides). Whereas the generation of coordinates seems > to produce an almost sensible structure: > [image: Inline images 3] > > subsequent geometry optimization does not: it moves the Eu atom way > outside of the coordination sphere: > [image: Inline images 4] > > Is it something with the bond types not specified correctly, or it is just > not supposed to work with this type of molecules at all? The molecule is > defined by the following SMILES (created with MarvinSketch): > > [Eu]1234567OC(=CC(=[O]1)C)C.C(C=C(O2)C)(=[O]3)C.C(C=C(O4)C)(=[O]5)C.C1=[N]6C2=C(C=C1)C=CC1=C2[N]7=CC=C1 > The same result is obtained with La instead of Eu. > Best wishes, > Michal > > > > -- > > ___ > 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] UFF geometry optimization of lanthanide complexes
Hello, I was trying to generate 3D coordinates for an europium complex, [Eu(acac)3(phen)], with UFF, using RDKit nodes in KNIME (UFF is parametrized for lanthanides). Whereas the generation of coordinates seems to produce an almost sensible structure: [image: Inline images 3] subsequent geometry optimization does not: it moves the Eu atom way outside of the coordination sphere: [image: Inline images 4] Is it something with the bond types not specified correctly, or it is just not supposed to work with this type of molecules at all? The molecule is defined by the following SMILES (created with MarvinSketch): [Eu]1234567OC(=CC(=[O]1)C)C.C(C=C(O2)C)(=[O]3)C.C(C=C(O4)C)(=[O]5)C.C1=[N]6C2=C(C=C1)C=CC1=C2[N]7=CC=C1 The same result is obtained with La instead of Eu. Best wishes, Michal -- ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss