Should reactions Smarts be aromatized ?

Good morning,





I have the following query.


We have a small set of Reactions which we use to enumerate Combinatorial Libraries.


So far, I have automatically aromatised the Smarts and used them directly.


However, a recent example has shown that this sometimes causes problem.





This is the example Smirks un-aromatised, the reaction produces the expected product.


select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6][N:2]1[N:3]=[N:4][C:5]([#1,#6])=[C:1]1[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y')


as product from dual;


CN1N=NC(C)=C1C





However, when using the aromatised Smirks, the product is different and a double-bond is missing.


select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6]-[c:5]1[n:4][n:3][n:2](-[#1,#6])[c:1]1-[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y')


as product


from dual


Cc1[nH][nH]n(C)c1C, which after de-aromatisation gives CN1NNC(C)=C1C (I have lost a double-bond in the product)





What do you recommend?


Aromatise the reagents but not aromatising the Reaction? My aim is to apply the same rules to all reactions if possible.





Thanks


Catherine

The examples from my previous post were running on 5.0.2





I have just done a quick comparison with 3.2.6 and I get the result below





select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6]-[c:5]1[n:4][n:3][n:2](-[#1,#6])[c:1]1-[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y')


as product


from dual


Product = Cc1nnn(C)c1C





select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6][N:2]1[N:3]=[N:4][C:5]([#1,#6])=[C:1]1[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y')


as product from dual;


Product = CN1N=NC(C)=C1C





With 3.2.6, the aromatisation of the reaction does not seem to make a difference.





Thanks


Catherine

Hi,

creisser wrote:

I have the following query. We have a small set of Reactions which we use to enumerate Combinatorial Libraries. So far, I have automatically aromatised the Smarts and used them directly. However, a recent example has shown that this sometimes causes problem. This is the example Smirks un-aromatised, the reaction produces the expected product. select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6][N:2]1[N:3]=[N:4][C:5]([#1,#6])=[C:1]1[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y') as product from dual; CN1N=NC(C)=C1C

Yes, this is the correct result.

Quote:

However, when using the aromatised Smirks, the product is different and a double-bond is missing. select jc_react('[#1,#6]-[#7-:2][N+:3]#[N:4].[#1,#6][C:5]#[C:1][#1,#6]>>[#1,#6]-[c:5]1[n:4][n:3][n:2](-[#1,#6])[c:1]1-[#1,#6]','C[N-][N+]#N.CC#CC','method:n reverse:n unambiguousReactionsOnly:n allowDuplProdLists:n ignoreErrors:y') as product from dual Cc1[nH][nH]n(C)c1C, which after de-aromatisation gives CN1NNC(C)=C1C (I have lost a double-bond in the product)

This result is not correct. The results should be the same whether the molecules in the reaction are aromatized or not. We will fix this bug in the next version.





Regards,


Zsolt

The problem seems to related to a bug in the handling of aromatic nitrogens in reactions. The problematic aromatized product SMILES contains implicit hydrogens, which is invalid in this case. Those hydrogens should not appear in the products. My colleagues will check the error and will get back to you soon.

Thanks


Catherine

The source of the error has been identified and corrected in the latest (5.0.4) JChem release.