Reaction failing to recognize substrates

We have written a reaction in the hopes that it would be able to predict products from caffeine and caffeine like substrates, among other things. We use jchem 3.2.3 and Marvin Sketch 4.1.5 but this reaction did not work with jchem 3.2.1 either. It cannot recognize caffeine (or theobromine, theophylline, paraxanthine, etc.) as a substrate. The attached marvin file (called bt0063) produces no products in Reactor and gives us the following result on the command line when caffeine is the substrate:

Code:

react -a changing -v -r ../molecules/bt0063.mrv "CN1c2ncn(C)c2C(=O)N(C)C1=O" Total running time (ms) : 296 Reaction setting (ms)   : 161 Reactant setting (ms)   : 1 Reaction processing (ms): 134

We would like the caffeine to be demethylated to produce: CN1c2ncn(C)c2C(=O)NC1=O


Is there a way we can map a general pattern so that it will predict the correct product for caffeine?


Thanks in advance!

The ring bonds in your reaction scheme are single aliphatic bonds, while the rings in your substrate are considered aromatic. If you replace the type of the corresponding bonds in the reaction scheme with aromatic or single-or-aromatic bond types, the reaction will work.

There are two rings in caffeine, one is five membered and one is six membered. Only the five membered ring is aromatic. The six membered ring CANNOT be aromatic, because this would create a pentavalent carbon. The SMILES strings of the substrate (caffeine) does reflect this. Also, the reaction scheme is written as an aliphatic ring because we want the methyl group on the six membered ring to be demethylated.

To give an update to this problem: after trying the suggestion given by the previous poster (make the bonds in the reaction "aromatic or single"), our system was able to predict products. However, all of the products had both rings as aromatic. This is incorrect. The six membered ring should be aliphatic and the five membered ring should be aromatic. As I stated previously, we are using Jchem 3.2.3 and Marvin Sketch 4.1.5. However, when I entered the reaction on the Reactor website and gave it caffeine as a substrate, the correct products appeared with the six membered ring being aliphatic and the five membered ring being aromatic. We are confused because we believed that Reactor also uses Marvin Sketch 4.1.5, so we are not sure why our products are different than yours.

Hi,

Rlong wrote:

To give an update to this problem: after trying the suggestion given by the previous poster (make the bonds in the reaction "aromatic or single"), our system was able to predict products. However, all of the products had both rings as aromatic. This is incorrect. The six membered ring should be aliphatic and the five membered ring should be aromatic.

We have two types of aromatization: "general" and "basic" (see the differences here). If caffeine is aromaized in general style, then also the six membered ring will be aromatic.





From JChem 3.2 the default aromatization type has changed to "general". When Reactor processes a reaction, then the input reactants are "slightly" standardized (=aromatized in "general" style) for search, and this causes the problem in this situation.

Quote:

As I stated previously, we are using Jchem 3.2.3 and Marvin Sketch 4.1.5. However, when I entered the reaction on the Reactor website and gave it caffeine as a substrate, the correct products appeared with the six membered ring being aliphatic and the five membered ring being aromatic. We are confused because we believed that Reactor also uses Marvin Sketch 4.1.5, so we are not sure why our products are different than yours.

I can not reproduce this. The JSP implementation of Reactor gives me the same results as the Reactor application (GUI or command line).


Reactor (and all JChem 3.2.3 applications) use Marvin 4.1.5.





Best regards,


Zsolt

Let me briefly explain the differences of the two aromatization methods. The basic method calculates the aromaticity of the given structure, the general method calculates the aromaticity of the given compound.





What is the difference?





Let's take 2-pyridone as an example (see the attached picture). The oxo form is certainly not aromatic. However, it has an tautomer called 2-pyridol (or 2-hydroxy-pyridine), which is aromatic. It dimerises in solution, but the oxo form is dominating over the OH form in the tautomeric equilibrium. Additionally, the oxo form has a mesomeric form, which is aromatic as well!





Thus, these kinds of compounds show aromatic features and can be considered aromatic.





I certainly understand, that drawing an aromatic circle in a ring having a doublebonded ligand hurts the eye of a chemist, but this is just a visualization issue.

If I can just put in my two cents here: I think most chemists would be surprised to hear the claim that the six-membered ring of caffeine (or pyridone, for that matter) is not aromatic. An aromatic compound has an odd number of electron pairs in a continuously overlapping cyclic array of p orbitals, and that is the case in both rings of caffeine. It is not necessary to have alternating pi bonds. The general method of aromatization that Gyuri describes definitely makes more chemical sense than the basic method.





Gyuri is also correct when he points out that representing the aromaticity of a compound like pyridone with a circle hurts the eyes of most chemists. (I call the circle representation the "toilet bowl" representation, both because of how it looks and because of where I would like it to go.) But this problem is not the fault of ChemAxon; Lewis structures are fundamentally inadequate at representing aromaticity, which can only be described adequately in molecular orbital terms. In other words, there is no good way to represent aromaticity in line drawings. Organic chemists just get used to seeing that certain compounds are aromatic without any visual representation of that property -- e.g., the 1,3,5-cyclohexatriene representation of benzene.

After reviewing this reaction further, we agree with the above. Both rings are and should be aromatic in both the substrate and products. Fixing this in our system will take some time and effort and we will begin working on this soon. Thank you all for your help! We greatly appreciate it.