Calculation result is not defined for query molecules

Hi,


I am trying to calculate properties from a XYZ file. Either marvinview or marvinsketch gives me this error "Calculation result is not defined for query molecules.", even when I tried to re-export it in an MRV format for example...


Attached is the xyz file


Cheers,


Nicolas

Hi Nicolas,





XYZ format does not store bond information, so the bonds are imported as ANY bond (see XYZ format help). ANY bond is a query bond type, the imported molecules are query molecules, that's why you get the error message "Calculation result is not defined for query molecules.". The re-exported molecules are also query molecules.





There is an "Any bond line style" option in MarvinSketch (Edit->Preferences->Bonds), please check this. By default this is set to "Automatic", I think that was the reason you haven't noticed ANY bonds.





You can use the "b" XYZ format import option (see XYZ import options) to guess the bond types, atom charges and implicit hydrogens from atom distances and valence rules (Warning: This is an experimental feature).





Regards,


Zsolt

Great, thanks!


I could calculate properties using

Code:

msketch test.xyz{b}

, not changing the msketch preferences, but it works - good to know about the ANY bond feature


Cheers,


Nicolas

Hi Zsolt,


thanks for your answer in my other topic.





I immediately tried msketch KFL-4K.xyz{b} , msketch opens, but there is no molecule in MarvinSketch, only an empty window. The command line does not show any error.


mketsch ethanol.mrv does not work also.


Is there any wrong setting?





Regards, Hans-Ulrich





I am using Marvin 5.0.3 on Linux.

Hi Nicolas,


which version of Marvin are you using ?


Regards, Hans-Ulrich

Hi Hans-Ulrich,


Looking at my post I may have been using version 5.0.2 or 5.0.1, if I can find the file I was working on I will check if it works with 5.0.3... Did you try to change the options to recognize any bonds?


Cheers,


Nicolas

I just tried with 5.0.3, I can open it with the {b} suffix, and I can calculate geometry property!

Hi saettel and totanec,





I just tried Marvin 5.0.1: It does not work also.





Are you using Windows, Mac or Linux ?





Regards, Hans-Ulrich

Hi,


Actually I am saettel and totanec (old username, sorry)


I am using WindowsXP, and it works either with the option "any bonds" set to automatic or single...


Cheers,


Nicolas

Hi Nicolas,





thank you. I am using Linux, and so it seems to be a problem of OpenSource or a problem of the JAVA library. I am using:





huw@tux1:~/Marvin> java -version


java version "1.4.2_05"


Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_05-b04)


Java HotSpot(TM) Client VM (build 1.4.2_05-b04, mixed mode)





Regards, Hans-Ulrich

Hi, I found it: With





java version "1.6.0_06"


Java(TM) SE Runtime Environment (build 1.6.0_06-b02)


Java HotSpot(TM) Client VM (build 10.0-b22, mixed mode, sharing)





it works. It's a good feature of Marvin !





Thank you for your help.





Regards, Hans-Ulrich

Hi,


the calculation of properties for query molecules is still blocked, please see the first screenshot. You showed a workaround (using the {b} option), but this means that I must start MarvinSketch each time I have a new query molecule. This is unhandy.





I have the idea to implement a new option in the call for msketch. There are options for starting msketch, please see the second screenshot.


So I suggest to have a new option, f.e. -b to "unblock" the calculation of properties for query molecules.





This could be done for desktop launching also, as I have shown in the topic http://www.chemaxon.com/forum/ftopic3852.html





Regards, Hans-Ulrich

Hi Hans-Ulrich,





I think you misunderstood something. The calculations can not be performed on query molecules, so using an option for unblocking the calculation of properties for query molecules is not a way to go. XYZ format has some constraints: it does not store bond types, and this is the problem. The {b} import option is a workaround to overcome the restrictions of the XYZ format.





I suggest the following: use molconvert to convert your molecules to a format that is capable of storing also bond data (e.g. smiles, sdf, mrv), and use the converted molecules for property calculations.





Example:

Code:

molconvert sdf test.xyz{b} -o test.sdf

In the example above the molecule in test.xyz file is converted to sdf; the result is stored in test.sdf (attached). The {b} XYZ format import option is also used, so query bond types are converted to "normal" bonds.





I hope this helps.





Regards,


Zsolt

Hi Zsolt,


before continuing the discussion, I must clear a problem I have with the test molecule, please see the screenshot. Which molecule is really meant ?


1) Left side: The molecule test.xyz , to which I have added the double bonds manually by analysing the topology of the carbon atoms.


2) Right side: The molecule test.sdf, to which molconvert has added two hydrogens. To this molecule it is impossible to add the double bonds without producing a biradical.





Hi Nicolas Saettel, which molecule is meant ?





Regards, Hans-Ulrich

Hi Hans-Ulrich,


Good point - the structure I meant was the aromatic one + 2 carbonyls instead of hydroxyls, so the one on the left where you manually added the double bonds.


Although I don't understand when you say that adding the double bonds creates a radical, msketch automatically transform OH in =O.


I don't know if there is an option that can aromatize and/or use double bonds when no explicit H are there, based on the geometry...


Cheers,


Nicolas

Hi Nicolas,





I didn't know the real structure as I was adding the double bonds to test.sdf. So I interpretated the hydroxygroups as belonging to enols and ended up with structure C :-(





Obviously there are to kinds of H-atoms seen while opening the test.sdf file: The one coming really from the xyz-structure and the one coming from the MarvinSketch option [ View-ImplicitHydrogens-OnHetero ], my normal setting.





Setting [ View-ImplicitHydrogens-Off ] and opening test.sdf you get the structure D and now it is clear how to put the double bonds.





My conclusions about this: I will put in the next post.





Cheers, Hans-Ulrich

right, D is the correct one (no implicit H) before adding the double bonds!

Hi, now I analysed the difference between the xyz-files and the by molconvert produced sdf-files, please see the screenshot for acetaldehyd (In the sdf-file there is obviously the date and time the file is created ;-).





The coordinates have now only 4 digits (why?) and there are lines with bond definitions below the coordinates. But there is no double bond between 2 and 3 !.





Opening either the xyz-file or the sdf-file leads to the same display. The convertion of the molecule does not work automatically and you have to add double bonds etc. manually.





So I suggest: Don't block the calculation of properties for xyz-input structures. The user has the responsibilty to add the correct bonds and to get the right results f.e. for the Dreiding Energy.





When the program molconvert does recognise the correct bond types by analysing the topology and the hybridisation of the atoms (i.e. total number of neighbors), then it can specify the bonds automatically.





Regards, Hans-Ulrich

Hi, I must give another example:





Let's assume, I've got the structure of the biocide MIT in xyz-Format and I don't know the structure. I transform it with:

Code:

molconvert sdf MIT.xyz{b} -o MIT.sdf

Then I open it with MarvinSketch (and the viewing option Implicit Hydrogen is accidentally "On all", which I don't remember). I get the structure seen in the screenshot.





This formula is not MIT, with serious errors when I would do some tool calculations.





MIT is Methylisothiazolinone, with a C=C and a C=O double bond.





Two consequences:


1) Either molconvert must produce multiple bonds


2) or MarvinSketch must automatically set "Implicit Hydrogen Off" when opening xyz- or sdf-files.





Regards, Hans-Ulrich

Hi,





Note: the "b" XYZ format import option (try to guess bond types) is an experimental feature. I mentioned this in my first post in this topic, but I haven't emphasized it enough, I think.





See: http://www.chemaxon.com/marvin/help/formats/xyz-doc.html





The relevant part:

Quote:

b Try to guess bond types, atom charges and implicit hydrogens from atom distances and valence rules. Warning: This is an experimental feature. Since XYZ files do not contain bond information, these "guesses" can be different from the actual bond orders, especially in case of - radicals - missing Hydrogen atoms - complexes The user should check the calculated bonds after import.

Regards,


Zsolt

Hi Zsolt,


yes, it's an experimental feature. And it is in the GUI also. To produce a structure format to which I can apply the calculation tools I evaluated the following way and documented the result by saving sdf-files.





1) After opening the xyz-file and saving the molecule as sdf, you get MITa.sdf and all bonds are undefined (value 8). The tools still cannot be applied.





2) Then using "Bond-Regenerate Bonds" with the options seen in the screenshot, and then saving as sdf, you get MITb.sdf and all bonds have the value 1 (single). Obviously did the "experimental" procedure not recognise the double bonds, as mentioned above. But now the tools can be applied ! (Of course in this example you get wrong results because of the missing double bonds.)





3) But having version b it is now easy to add the double bonds. And saving leads to MITc.sdf with correct bonds. And now you can apply all tools !





OK, this is a way to do it in courses, but it should go easier.





Step 2) corresponds to the command line import with option {b}. So the GUI should do the import with this option automatically (and with "Implicit Hydrogen = Off" ), why not ?. Then the user can check the structure, correct it, save it and apply the tools.





Regards, Hans-Ulrich

Hi,

HUWagner wrote:

Step 2) corresponds to the command line import with option {b}. So the GUI should do the import with this option automatically (and with "Implicit Hydrogen = Off" ), why not ?. Then the user can check the structure, correct it...

... or start sending bug reports because single bonds are imported instead of double bonds, some bonds are not present, other imported bonds are really non-existent etc. So the {b} option must never be default.





We will consider the possibilities to make import option setting easier in the GUI.





best regards,


Peter

Hi,





the problem with the setting "Implicit Hydrogen On All" still exists in MarvinSketch 5.1.1. To show it once more, I append the screenshot bf0085. MIT.xyz is converted to MIT.sdf using the command line procedure molconvert sdf MIT.xyz{b} -o MIT.sdf, mentioned in earlier posts in this topic.





This can produce accidentally problems to non experienced users of MarvinSketch. I suggest once more: The option "Implicit Hydrogen On All" should be set to "Off" automatically before importing those molecule files. When you do not remember that you has set it "On", then you get eventually wrong results. I think it is not a problem of convenience. At least there should be a warning.





Regards, Hans-Ulrich





P.S. This post does not discuss the experimentally option {b} mentioned in the posts above.

Hi,

HUWagner wrote:

The option "Implicit Hydrogen On All" should be set to "Off" automatically before importing those molecule files.

Besides the fact that presently there is no mechanism in MarvinSketch to change the View menu settings based on file types: how could MarvinSketch find out that an SDFile was created from an XYZ file format by molconvert? Do you think we should store a history of conversions in the file?





Best regards,





Akos

Hi Akos,





thank you for your answer. No, the history should not be stored in the files.


I will look for another possibility to solve the problem.





Regards, Hans-Ulrich