Clean 2D

Hi,





When I paste in O=C(NC1=CC=CC=C1N1CCNCC1)NC1=CC=CC=C1C1=CC=CC=C1 smiles into MarvinSketch 5.0.3, it looks ugly.





Can this be cleaned?





thanks,

Thanks for the report.


I agree that it is ugly, but I'll fix it ASAP.


Marvin 5.1 is in testing phase so the fix will appear in 5.1.1.





Andras

Thanks! Do you have an estimate release date for 5.1.1?

One month later than 5.1.

Hi,





Is this still on track?





Also, when I pasted CCCN1CCC2=C(C1)SC(NC(=O)CCS(=O)(=O)C1=CC=C(C)C=C1)=C2C(N)=O into MarvinSketch 5.1.0, the two carbonyls are most touching. Can this be cleaned?





thanks,

Hi,





The original molecule is cleaned properly in 5.1.1, which is going to be released these days. On the other hand, in your last example the two carbonyls are still very closed to each other; Andras is on holiday now, but he will investigate this problem as soon as he is back.





Best regards,





Akos

Thanks!





One thing I want to mention is that sometimes when you tried to improve "clean 2D" algorithm, the "old clean" molecules looked ugly when new algorithm is applied. So I think regression testing is very important.

Hi,





It was just a luck that some modification in the clean 2D algorithm is solved one problematic molecule you have mentioned. Actually the fix is due in Marvin 5.1.1.


Thank you for calling our attention to regression tests. Actually we are continuously collecting molecules from the forum to the test set, which is checked before each release.





Andras

Hi,





I just tried Marvin 5.1.1. It cleaned the original molecule/smiles very well (Thanks!), but for CCCN1CCC2=C(C1)SC(NC(=O)CCS(=O)(=O)C1=CC=C(C)C=C1)=C2C(N)=O, the two carbonyls are still very close...





Will the fix be included in version 5.1.2?





Thanks,

Hi,





The problem is the selection of the allowed minimum distance between two atoms (collision_distance). Actually, if the distance is smaller than this given value, then the algorithm tries to find an other solution by rearranging bonds.


In case of this molecule it is obvious that there exists a better solution (even if this collision_distance is larger). But this is not always the case. So I think the solution would be to allow the user to set this collision_distance by giving it as parameter for the algorithm.





What do you think?





Andras

I think it should be adjusted by the clean algorithm automatically. If the user sets collision_distance, it might break other molecule's display.





Attached is the comparison result between Marvin 5.1.1 and Chemdraw.





Thanks!

Thank you for the attachment.


We will improve it by 5.1.2.


Andras

Great! Thanks!

Marvin 5.1.2 is out.


I hope the result as you needed.

It looks great!





Many thanks!