Conformers and Hydrogen bonds

Hi,


I have a question about hydrogen bond treatment in MarvinView. I am trying to generate conformers of a complex where the two components are only potentially bonded by hydrogen bonds and/or stacking...


The conformer generation plugin separates the 2 molecules and generates independant conformers (where intramolecular hydrogen bonds can be present, I notice). Is it possible to specify hydrogen bond constraints?


Thanks


Nicolas

Hi,





No, intermolecular hydrogen bonds are not handled by conformer plugin.





Regards,


Zsolt

Thanks for the answer... is this possible and/or envisionned?


Cheers,


Nicolas

saettel wrote:

Thanks for the answer... is this possible and/or envisionned? Cheers, Nicolas

Hi Nicolas,





The only problem is that currently intrermolecular H-bonds are not treated as "bonds", therefore, the parts connected with only H-bonds are treated as separate fragments and arrive to the conformer generation process separately. The explicite treatment of H-bonds is planned but should be solved at multiple levels as it effects the behavior of the fragmentation of a "Molecule". Unfortunately, I cannot tell you an explicite date now when this task is expected to be complete.





Best wishes,





Ödön

Hi,

Have you thought more about how to represent a hydrogen bond in MarvinSketch?  Any plans to introduce such a bond?

-- Bob

Dear Bob,

We have not considered to support such bond until you convince us to consider.

Andras

Well, a hydrogen bond is certainly a very important kind of bond in organic chemistry, and for some applications it's absolutely essential to describing structure and behavior.  One of our users is using dative bonds as a workaround for the lack of H bonds, but obviously that's not a satisfactory solution.

My guess is that your customers in the pharmaceutical industry would also very much appreciate the ability to indicate H bonds in their structures.

Further up in this discussion, you said that the main factor holding you back was how to fragment a Molecule containing entities held together by H bonds.  I don't think this is a very difficult problem to address.  By default, you break H bonds when fragmenting a Molecule, but you add an option for leaving them intact.

bobgr wrote:

Further up in this discussion, you said that the main factor holding you back was how to fragment a Molecule containing entities held together by H bonds.  I don't think this is a very difficult problem to address.  By default, you break H bonds when fragmenting a Molecule, but you add an option for leaving them intact.

I don't think fragmentation would be a problem. We have already multicenter bond which can be excluded optionally during fragmentation.

So actually Hydrogen bond development is not in high priority. But we will definitely consider.

I have recorded this feature request.

Hi,

to use "dative" bonds for hydrogen bonds is not a workaround, it is absolutely correct.

The Marvin bond menu uses for "dative" bonds the synonym "coordinate". And using this bond type to build the important examples in the upper screenshot produces exact hydrogen bonded dimers.

The hydrogen bond is not an covalent bond. Besides the charge attraction there is a two electron bond with two electrons coming from the bond donor (here N: or O:) and the hydrogen is the acceptor. This part corresponds to normal "dative = coordinate" bonds.

Drawing the hydrogen bonds from the donor atom (here O: and N:) produces the shown bond arrows and one electron pair vanishes. These are very good formulas.

Using the tool "Conformers" produces the molecule pairs shown in the second screenshot. This is exactly the request Nicolas Saettel posted here in Oct. 2007.

MarvinSketch is excellent in handling hydrogen bonds.

There may be one "design" change/addition: Hydrogen bonds could be shown as dotted arrows. But the arrow must be conserved: The hydrogen bond has obviously a direction.

Regards, Hans-Ulrich

HUWagner wrote:

to use "dative" bonds for hydrogen bonds is not a workaround, it is absolutely correct.

We're getting into the realm of conventional representations here in addition to chemistry.

Dative bonds are covalent bonds.  For example, the P-Pd bonds in (Ph₃P)₂PdCl₂ are typical dative bonds.  I show three ways of depicting this compound below.  The first depiction, which shows the formal charges, is not used by anyone.  The second depiction, which uses dative bonds and omits formal charges, is perfectly correct, but no one uses straight arrows to show dative bonds anymore.  The third depiction, which uses regular bonds and omits formal charges, is what is used most widely today.  Nevertheless, the first and second depictions are the only choices in MarvinSketch.  I could not use MarvinSketch to draw the third depiction; I had to use ChemDraw.  

By contrast, hydrogen bonds are not covalent bonds; they mostly involve electrostatic interactions, and they are substantially weaker than most covalent bonds.   In modern chemical literature, hydrogen bonds are represented by dotted, dashed, or hashed lines.  I have never seen them represented by the straight arrow that represents dative bonds, let alone by a dashed or hashed arrow.

I agree that the straight-arrow bond symbols in MarvinSketch can be used to represent H bonds, but I disagree that it is good practice to do so, because it teaches students a way of representing these bonds that is not conventional.

Other problems if JChem does not understand the difference between dative bonds and hydrogen bonds:

1. Certain calculations (bond lengths, energies, etc.) may not be valid.


2. It should be possible to fragment just the H bonds in a structure to separate components of a hydrogen-bond mixture or to minimize a structure's conformation in the absence of H bonds.

My suggestions are as follows:

1. Provide an option in MarvinSketch for depicting the dative bond as a regular single bond (without the arrow head).  It should be possible for both the user and the JSP page to set this option.  This option will allow users to draw (Ph₃P)₂PdCl₂ in the third depiction.


2. Provide a way of depicting a hydrogen bond in MarvinSketch, and distinguish its internal representation (bond type) from a dative bond, so that a user or programmer can treat these types of bonds in different ways.  


3. Provide a Java method for breaking all H bonds in a Molecule.  With this options, it will not be necessary to add an option to convertToFrags() method; the method should leave the H bonds intact.

Hi,

The arrow has one more meaning in the "dative" bond representation. The tip of the arrow always must point to an unoccupied  vacancy in the valence shell of a central atom. E.g. "d" or "f"  shell of metal atoms play this role.     

On the other hand the tip of the arrow in case of  "H" atom does not have above meaning. The "dative" bond arrow symbol may be confusing in case of H-bond.  

Jozsi

Jozsi wrote:

Hi, The arrow has one more meaning in the "dative" bond representation. The tip of the arrow always must point to an unoccupied  vacancy in the valence shell of a central atom. E.g. "d" or "f"  shell of metal atoms play this role.      On the other hand the tip of the arrow in case of  "H" atom does not have above meaning. The "dative" bond arrow symbol may be confusing in case of H-bond.   Jozsi

Yes, exactly, that's what I meant when I said that the dative bond is covalent, whereas the hydrogen bond is electrostatic.

In reality, there's no chemical difference between a dative bond and a single bond.  The dative bond representation just allows you to avoid drawing the formal charges.  In the context of MarvinSketch, a dative bond keeps valence errors from appearing when you omit formal charges.  But it would be useful for the user to be able to make the dative bond look like a single bond.

The colleague who asked me about hydrogen bonds in MarvinSketch has now asked me about drawing partial bonds, as in a transition state.  I told her that the ANY bond looks like a partial bond in a transition state drawing, so she could use that, but using an ANY bond to represent a partial bond is even more problematic than using a dative bond to represent a hydrogen bond, because of how JChem treats the ANY bond in structure comparisons.  So, I would also like to request that you add a "partial" bond to MarvinSketch.  Give it a unique bond type value, and treat it as different from single, double, and triple.  

Hi Bob and Jozsi, in consequence of your detailed descriptions I will change my sentence above to:

It is a workaround to use dative bonds for hydrogen bonds. These different bond types should have different handling and display.

And I agree Bob's suggestions. Only one question: "How to display the hydrogen bond on the canvas?" The literature mainly uses dotted lines, but these are in MarvinSketch used by "Any".

Best regards, Hans-Ulrich

The ANY bond consists of small dots, like so: .......

For a hydrogen bond, MarvinSketch could use larger dots, like so: •••••••

And for a partial bond, it could use a dashed bond, like so:   - - - - - 

There is still a possibility of confusion between the ANY bond and the partial bond, but I imagine that any user knowledgeable enough to draw an ANY bond will be knowledgeable enough not be confused by their similar appearances.

Still, I would suggest that a tool tip should appear when the user hovers over an ANY, hydrogen, or partial bond, to help avoid confusion.

-- Bob

Hi Bob and Hans-Ulrich,

- We already have a feature request in our issue tracking system for the option to display dative bond as a single bond. I increase its priority now, so it will be supported soon.

- Regarding the new bond types: I agree with most of your ideas (although the display should be discussed in more detail), and we can certainly support these bond types in MRV and CXSMILES formats, but we have to think how they will be exported to other file formats (SMILES, MOL, SKC, CDX). Also, we should be sure that we follow IUPAC recommendations with the representation. If there is no recommendations yet, we could initiate to create the standards for these bond types, and then the other programs/fileformats will follow :)

Best regards,

Akos 

akospapp wrote:

We already have a feature request in our issue tracking system for the option to display dative bond as a single bond. I increase its priority now, so it will be supported soon.

Thanks.  As usual, please make it possible to set this value from either the GUI or from Javascript.

I have recorded the request for two new bond types: hydrogen bond, partial bond.