Aromaticity? - Call for opinions

It is a common and interesting question; when is a molecule aromatic and how. A related question, is the level of aromaticity.





Since there is ambiguity in defining this issue I thought it would be a good idea to begin a discussion on the topic.





To start this off I post 6 molecules in Kekule form:

1. pyridin-2(1H)-one O=C1NC=CC=C1
2. 2-thioxo-2,3-dihydropyrimidin-4(1H)-one O=C1NC(=S)NC=C1
3. 2,4-dihydro-3H-1,2,4-triazol-3-thione S=C1NC=NN1
4. 9H-xanthen-9-one O=C1C2=CC=CC=C2OC3=C1C=CC=C3
5. 5-thioxo-1,2,4-triazolidin-3-one O=C1NNC(=S)N1
6. imidazo[1,5-a]pyridine-3(2H)-thione S=C1NC=C2C=CC=CN12
   
   
   

For me, starting from left to right their status is:

1. aromatic
2. questionable
3. questionable
4. the middle ring is not aromatic
5. not aromatic
6. questionable

What do you think?





Andras

IMHO, all of the compounds you drew have some aromatic character. All are lower in energy than would be expected from the corresponding acyclic compounds, and all would probably show particularly high deshielding in their NMR spectra.





The problem with defining whether Lewis structures are aromatic is that aromaticity is fundamentally a molecular orbital phenomenon. Much of organic chemistry can be described quite well with Lewis structures, but aromaticity cannot. In fact, the explanation of the unusual stability of benzene was probably the first MO argument that was recognized by organic chemists as having some relevance to their field.

Do you think we should handle these structures aromatic or non-aromatic in structure searching? The cd_smiles column of JChem Base tables stores aromatic structures in aromatic form. We are not sure if these molecules should be stored with aromatic bonds or as they are displayed in the image of Volfi. Should a substructure containing aromatic bonds provide hits for these structures? Or it is better if the user has to draw the substructure with single/double bonds to get hits?


For example searching with O=cncc or O=CNC=C should result the first structure as a hit?

Ferenc wrote:

Do you think we should handle these structures aromatic or non-aromatic in structure searching? The cd_smiles column of JChem Base tables stores aromatic structures in aromatic form. We are not sure if these molecules should be stored with aromatic bonds or as they are displayed in the image of Volfi. Should a substructure containing aromatic bonds provide hits for these structures? Or it is better if the user has to draw the substructure with single/double bonds to get hits? For example searching with O=cncc or O=CNC=C should result the first structure as a hit?

I think we discussed this subject before. All aromatic resonance forms should be considered in any search. It is counterintuitive that it should be necessary to aromatize a structure before searching, and it is counterintuitive that only compounds that are in the database in aromatized form should match it. In other words, the default should always be aromatization. Perhaps, in the rare cases that aromatization is not desired, an option for nonaromatization could be provided.





-- Bob

bobgr wrote:

All aromatic resonance forms should be considered in any search.

If we would like to avoid serious speed problems, this requires a modified database structure. Presently, a structure table contains one row for each molecule. One solution is that another table is created just for resonant molecules. This table would store the alternative fingerprints and structures (cd_fp<n> and cd_smiles colums) of molecules having different resonance forms. We need to maintain compatibility with previous versions, but I think we can implement something like this, though it is a major change and requires time.


Ferenc





PS:


The rest of your comments are not strictly related to this issue. Hi level functions, like JChemSearch do consider aromaticity, but low levels, like MolSearch require aromatic conversion before usage. Your suggestion has advantages and disadvantages, but I suggest to discuss them in another topic.

Hello everybody

We are currently dealing with a lot of structures where we want to determine aromaticity. I know that it is difficult to write an algorithm that is accurate, especially for heterocyclic systems. Electron drawing groups, lone pairs and many other factors influence whether a ring system is aromatic or not.

The 2n+2 rule based daylight algorithm is quite efficent respective to its simplicity. We would like to add one simple rule to reduce the number of "false positives":

if (number of pi electrons in ring > size of ring + 1)

    aromatic = false;

This prevents for example 6-membered rings with 10-pi electrons to be aromatic but also ensures that 5- or 7-membered rings can be aromatic.

See some examples in the attached picture.

We are grateful for any comments and suggestions about this idea.
Lorenz & the Reymond Group

Your idea seems to be correct and I like it very much.

The only problem is how can one add this simple rule to the daylight algorithm.

The problem is as follows:

The algorithm uses ring systems and not rings. The atoms in the ring system
may not form a continuous ring like in benzo[de]chromene
(SMILES: O1C=CC2=CC=CC3=CC=CC1=C23)
where all atoms (including the center atom) of the molecule are in the aromatic ring system generated from the
three 6 membered ring.

So in case of porphirine the rings are not aromatic alone (except two five membred rings) but the whole ring system is aromatic having 26 electrons and 24 atoms.

Using the given rule the porphirine whould have only two 5 membered aromatic ring, which is not true.

Andras

 

 

Thanks for the suggestion about the prophyrin. We are currently checking it.

Meanwhile we have found another rule:

if(ring has exocyclic C=C bond)
aromatic = false;

The reason for this is a very famous example: Ortho-Quinodimethane (C=C1C=CC=CC1=C). It is known that this is a reaction intermediate which has the tendency to create cycles to gain back aromaticity but the structure itself is not aromatic.

Lorenz

OK, this should be it. My labmate Ruud van Deursen coded around half a day, the result is attached. The code is based on jchem 5.2.6. There is also a picture showing which structures are aromatized and which not.

Please check the code. We believe it is already much better than daylight...

Lorenz

Summary:

aromatize according to huckel's ruel.
if(ring has exocyclic C=C bond)
aromatic = false;
if(number of pi electrons in ring > size of ring + n)
aromatic = false;

Great!

I will definitely check it asap.

Andras

Hi!

It is very nice indeed.

I have found a counter example:

9H-xanthen-9-one O=C1C2=CC=CC=C2OC3=C1C=CC=C3

The middle ring is not aromatic.

An other thing:

calling mol.getAromaticAndAliphaticRings(..) for fullerenes is like an endless loop as these molecules has so many rings... (I have not calculated yet theoretically)

I'm still checking.

Andras

Hello

Are you sure that the middle ring in xanthen-9-one is not aromatic? We have found literature (cristallographic) data which show that it is a overall planar, aromatic structure...

Well even if you are right: we do not claim that our aromatizer is perfect. :) The goal of it is to have a simple aromatizer (like the original one from daylight) where the results can be unterstood and checked by hand if necessary. So we don't like "fuzzy" algorithms with a lot of exceptions and special cases but prefer something simple that is not perfect but we immediately understand the results. Because then you know exactly with what you are dealing.

Therefore my coallegue took out the ChargePlugin from the code and replaced it by simple, empirical rules. It also makes it lots faster. Please have a look at the new code and check back on the xanthen-9-one example (do you have more?).

Lorenz Blum and the Reymond Group

Hi,

Therefore my coallegue took out the ChargePlugin from the code and replaced it by simple, empirical rules...

 

Is there any relaion between the ChargePlugin and the aromatization?

 

 

Jozsi

jszegezdi wrote:

Is there any relaion between the ChargePlugin and the aromatization? Jozsi

We had the ChargePlugin in for determining whether an atom on the ring was electron poor or electron rich.

We replaced now this function with empirical rules. This makes the aromatization faster and the results are easier understandable.

lorenz

Hi, I have a question here:

In which context do I need the property "aromatic" or "not aromatic" ? What is the purpose to estimate this property ?

In the original definition it was a model to explain the special stabiltiy and "non-reactivity" of benzene and those molecules.

Regards, Hans-Ulrich

Hi,

For structure search you will definitely need it.

Otherwise the two molecule would not be equal (note the rearranged double bonds), but they are.

Andras

Hi,

OK, I agree for molecules like this ortho-xylene.

But I will look for examples which are not "aromatic" and have the same problem.

Hans-Ulrich

What you are really doing with xylene is saying that the two resonance structures represent the same compound.  The fact that xylene is aromatic is irrelevant; the transformation of the alternating bonds into "aromatic" bonds is simply a shorthand way to transform both resonance forms into a single structure.  

It is also true that, e.g., [CH2+][O-] and [CH2]=O represent the same compound.  Here, though, one needs to use a different method to convert both structures into a single one: generation of a "canonical" resonance structure.  (We haven't used this method, so I don't know how well it works.  Instead, we find that two structures are resonance structures if and only if they have identical sigma-bond networks and identical overall charges and do not satisfy JChem's perfect match algorithm.)  

The difference between the two cases, of course, is that the two xylene resonance structures are nearly equal in energy, and there is no preference for drawing xylene in one or the other resonance form, whereas these things are not true of the two formaldehyde structures.

Happy New Year.

Hi Bob, one comment and one question:

1) " ... the transformation of the alternating bonds
into "aromatic" bonds is simply a shorthand way to transform both
resonance forms into a single structure.  " . Yes, but it is once more wrong nomenclature: Call it "delocalisation of bonds" and "delocalised bonds". Many, many users, especially when using the internet-Marvin-applets, learn this wrong nomenclature, f.e. using the buttons "Basic-" "General-" or "Loose"-Aromatization.

2) "... sigma-bond networks ". Do you mean sigma- or pi-bond networks ?

Happy New Year.

Hans-Ulrich

I agree that "delocalized" (American spelling) or "delocalised" (British spelling) is better than "aromatic".

No, I meant sigma-bond network.  [CH2]=O and [CH2+][O-] do not have the same network of pi bonds, but they are resonance structures.

My definition of resonance structures is not without its ambiguities; some people would call [CH2+][CH3] and [CH2]=CH2].[H+] resonance structures if the distance between C and H+ in the latter structure was unchanged from the former structure.  But because SMILES, at least, has no spatial information, I think we must treat this case as not resonance structures.  

Hi Andras,
as you can see in the image, I have analysed an example having problems in determining "aromaticity".
And I have two points:

1) I showed the formula G1 to organic chemistry colleagues, and they all commented "formula G1 is inacceptable".
And I repeat: Many, many users, especially when using the internet-Marvin-applets, learn this wrong nomenclature.
May be those formulas are handy for use in chemical databases (SMILES-strings) but these are bad for use in education.

The option "GENERAL" is the default option.
And when you change it to "BASIC", the next time you call MarvinSketch (applet or installed),
MarvinSketch does not remember the change. So I have the demand:

Make the aromaticity option "BASIC" to the default option, please as fast as possible.

The desired "remember function" could be installed later, but it would be nice to have it for the database user.

2) The image shows also the Marvin handling of the mesomeric resonance structures R1<->R2.
Here the assignment to aromaticity corresponds to the formula R2.

The Marvin handling of the dipolar resonance formula R2 is different:
2a) F.e. the Huckel HMO is now able to handle the pi-system R1<->R2 exactly as one structure.
2b) F.e. the logP tool uses the canonical form (here R1) and calculates the properties for this formula. To use the canonical form is a good workaround.
2c) F.e. the pKa tool results in totally different values for the two mesomeric resonance formulas. This is inacceptable and must be corrected.
But these mesomeric resonance problems may lead to a new topic with the title "Mesomerism / Resonance: Call for opinions".

So I will repeat the demand given above:

Please change the aromaticity default option to "Aromatize BASIC" .

So the "normal" user of MervinSketch will get unproblematic BASIC aromaticity results, and the user who needs more advanced SMILES strings can change the option to "GENERAL".

Best regards, Hans-Ulrich

Dear Hans-Ulrich,

Thank you for your comments.

In 5.3.1 the names will change in the GUI:

Aromatization -> Aromatic form

Aromatize -> Convert to aromatic form

Aromatization method -> Conversion method

Dearomatize -> Convert to Kekule form

The selected conversion method will be saved during exit and the default conversion method in the GUI will be the BASIC method.

Andras

Dear Andras, SUPER !   Best regards,   Hans-Ulrich

Hi Hans-Ulrich,

I would like to inform you that we had to make an urgent patch release that was named 5.3.1, therefore the renaming will be included only in the patch release coming at the end of March (probably named 5.3.2).

Best regards,

Akos

Hi Akos, super !

As shown in the attached image, now we have the new Tool Structure -> Aromatic Form -> Convert to ...

and the structure is shown with delocalized bonds,

and the options you choose "Basic - General - Loose" are saved while closing MarvinSketch and when you start it once more you have the same option as before.

Thank you. Best regards, Hans-Ulrich