cdk-devel — Developers forum for discussion on the Chemistry Development Kit

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Egon W. <ego...@gm...>]

https://github.com/cdk/nodes4knime/tree/master/org.openscience.cdk.knime/src/ambit2/knime/tautomers

On Fri, Dec 21, 2018 at 11:06 AM Nina Jeliazkova <jel...@gm...>
wrote:

> Dear All,
>
> Could you please point to the code of KNIME node for ambit-tautomers?   We
> could help upgrading (and enhancing with more options) to the newest
> version.
>
> We have up-to-date branch
> https://sourceforge.net/p/ambit/git/ci/ambit4/tree/ with cdk2 dependency,
> which is going to be merged with master in few days.
>
> Best regards,
> Nina
>
>
> On Mon, 12 Mar 2018 at 08:17, Nina Jeliazkova <jel...@gm...>
> wrote:
>
>>
>>
>> On 11 March 2018 at 18:41, Egon Willighagen <ego...@gm...>
>> wrote:
>>
>>>
>>>
>>> On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
>>> chr...@un...> wrote:
>>>
>>>> With regard to your statement below: Is there a chance to disentangle
>>>> the actual CDK nodes from JCP and AMBIT?
>>>> For JCP, there is the excellent Marvin Sketch node as a temporary
>>>> replacement.
>>>>
>>>
>>> That sounds like a reasonable back up...
>>>
>>>
>>>> Not sure if the AMBIT developers can help migrate to 2.x or move to a
>>>> dedicated repository?
>>>>
>>>
>>> I think we can get AMBIT to work... I will coordinate this with Nina.
>>>
>>
>> Yes, we already have a branch with CDK2 dependency
>> <http://svn.code.sf.net/p/ambit/code/branches/ambit-cdk-2/> , there are
>> some failing tests, mostly related to stereochemistry.
>>
>> Regards,
>> Nina
>>
>>
>>>
>>> Egon
>>>
>>>
>>> --
>>> E.L. Willighagen
>>> Department of Bioinformatics - BiGCaT
>>> Maastricht University (http://www.bigcat.unimaas.nl/)
>>> Homepage: http://egonw.github.com/
>>> LinkedIn: http://se.linkedin.com/in/egonw
>>> Blog: http://chem-bla-ics.blogspot.com/
>>> PubList: https://www.zotero.org/egonw
>>> ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
>>> ImpactStory: https://impactstory.org/u/egonwillighagen
>>>
>>
>>

-- 
Hi, do you like citation networks? Already 51% of all citations are
available <https://i4oc.org/> available for innovative new uses
<https://twitter.com/hashtag/acs2ioc>. Join my in asking the American
Chemical Society to join the Initiative for Open Citations too
<https://www.change.org/p/asking-the-american-chemical-society-to-join-the-initiative-for-open-citations>.
SpringerNature,
the RSC and many others already did <https://i4oc.org/#publishers>.

-----
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
Blog: http://chem-bla-ics.blogspot.com/
PubList: https://www.zotero.org/egonw
ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
ImpactStory: https://impactstory.org/u/egonwillighagen

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Nina J. <jel...@gm...>]

Dear All,

Could you please point to the code of KNIME node for ambit-tautomers?   We
could help upgrading (and enhancing with more options) to the newest
version.

We have up-to-date branch https://sourceforge.net/p/ambit/git/ci/ambit4/tree/
with cdk2 dependency, which is going to be merged with master in few days.

Best regards,
Nina


On Mon, 12 Mar 2018 at 08:17, Nina Jeliazkova <jel...@gm...>
wrote:

>
>
> On 11 March 2018 at 18:41, Egon Willighagen <ego...@gm...>
> wrote:
>
>>
>>
>> On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
>> chr...@un...> wrote:
>>
>>> With regard to your statement below: Is there a chance to disentangle
>>> the actual CDK nodes from JCP and AMBIT?
>>> For JCP, there is the excellent Marvin Sketch node as a temporary
>>> replacement.
>>>
>>
>> That sounds like a reasonable back up...
>>
>>
>>> Not sure if the AMBIT developers can help migrate to 2.x or move to a
>>> dedicated repository?
>>>
>>
>> I think we can get AMBIT to work... I will coordinate this with Nina.
>>
>
> Yes, we already have a branch with CDK2 dependency
> <http://svn.code.sf.net/p/ambit/code/branches/ambit-cdk-2/> , there are
> some failing tests, mostly related to stereochemistry.
>
> Regards,
> Nina
>
>
>>
>> Egon
>>
>>
>> --
>> E.L. Willighagen
>> Department of Bioinformatics - BiGCaT
>> Maastricht University (http://www.bigcat.unimaas.nl/)
>> Homepage: http://egonw.github.com/
>> LinkedIn: http://se.linkedin.com/in/egonw
>> Blog: http://chem-bla-ics.blogspot.com/
>> PubList: https://www.zotero.org/egonw
>> ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
>> ImpactStory: https://impactstory.org/u/egonwillighagen
>>
>
>

[Cdk-devel] Get java.lang.NullPointerException when call getAllSurfacePoints()

[Qinqing L. <liu...@gm...>]

Hi,

I am new to CDK.
I just followed an example in groovy, and I want to read a pdf file, get the points on molecule surface.

The code is below:
PDBReader reader = new PDBReader(new FileInputStream(file)); 

ChemFile crambin = reader.read(new ChemFile()); 

List<IAtomContainer> containers= ChemFileManipulator.getAllAtomContainers(crambin); 

for (int i = 0; i< containers.size(); i++) { 

	IAtomContainer container = containers.get(i); 

	String description = container.toString(); 

	NumericalSurface new_surface = new NumericalSurface(container); 

	Point3d[] points= new_surface.getAllSurfacePoints(); 

}
catch the exception java.lang.NullPointerException in last step.

Anyone can help me with this?
Thanks a lot!

[Cdk-devel] Fwd: Hydrogen Distribution

[John M. <joh...@gm...>]

- John

Begin forwarded message:

> From: Mehmet Aziz YİRİK <meh...@ou...>
> Date: 1 November 2018 at 14:46:09 GMT
> To: John Mayfield <joh...@gm...>
> Subject: Re: [Cdk-devel] Hydrogen Distribution
> 
> Dear Mr. Mayfield,
> 
> Thanks a lot for your recommendations. I have modified my class but have a question. You said "likely wants to take IMolecularFormula rather than a string." Do you mean as an input for HydrogenDistributor taking the IMolecularFormula or for all the functions of the class, rather than string  IMolecularFormula should be used ? For some of them it is easier to take string as an input. 
> 
> Kind Regards,
> Aziz
> From: John Mayfield <joh...@gm...>
> Sent: 29 October 2018 20:23
> To: Developers forum for discussion about the Chemistry Development Kit (CDK)
> Subject: Re: [Cdk-devel] Hydrogen Distribution
>  
> Sounds interesting, org.openscience.cdk.tools.CDKHydrogenAdder is not the right place though but possible in the 'cdk-structgen' module. The single class you have looks good just needs to be library-fied (e.g. remove PrintWriter/commons options etc). Also likely wants to take IMolecularFormula rather than a string.
> 
> p.s. you should use atomic number rather than atom symbol in your Map. At the moment you couldn't handle 'Na' for example.
> 
> On Mon, 29 Oct 2018 at 16:13, Mehmet Aziz YIRIK <meh...@ou...> wrote:
> Dear CDK Developers,
> 
> I developed a helper class called HydrogenDistributor. This class simply takes a list of atom symbols  and the number of hydrogens as inputs. The class calculates combinatorially all the possible distributions of these hydrogens. This class has been developed first as a helper class for a structure generator but can be also added to org.openscience.cdk.tools.CDKHydrogenAdder package, if you also think that should be under this CDK package. 
> 
> The class is shared on GitHub with its test class as a maven project.
> HydrogenDistributor: https://github.com/MehmetAzizYirik/HydrogenDistributor
> 
> Please let me know if it can be added to this CDK package (or any other CDK packages) or not. 
> 
> Kind Regards,
> Mehmet Aziz Yirik
> 
> 
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel

Re: [Cdk-devel] Hydrogen Distribution

[John M. <joh...@gm...>]

Sounds interesting, org.openscience.cdk.tools.CDKHydrogenAdder is not the
right place though but possible in the 'cdk-structgen' module. The single
class you have looks good just needs to be library-fied (e.g. remove
PrintWriter/commons options etc). Also likely wants to take
IMolecularFormula rather than a string.

p.s. you should use atomic number rather than atom symbol in your Map. At
the moment you couldn't handle 'Na' for example.

On Mon, 29 Oct 2018 at 16:13, Mehmet Aziz YIRIK <meh...@ou...>
wrote:

> Dear CDK Developers,
>
> I developed a helper class called HydrogenDistributor. This class simply
> takes a list of atom symbols  and the number of hydrogens as inputs. The
> class calculates combinatorially all the possible distributions of these
> hydrogens. This class has been developed first as a helper class for a
> structure generator but can be also added to org.openscience.cdk.tools.CDKHydrogenAdder
> package, if you also think that should be under this CDK package.
>
> The class is shared on GitHub with its test class as a maven project.
> HydrogenDistributor:
> https://github.com/MehmetAzizYirik/HydrogenDistributor
>
> Please let me know if it can be added to this CDK package (or any other
> CDK packages) or not.
>
> Kind Regards,
> Mehmet Aziz Yirik
>
>
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>

[Cdk-devel] Hydrogen Distribution

[Mehmet A. Y. <meh...@ou...>]

Dear CDK Developers,

I developed a helper class called HydrogenDistributor. This class simply takes a list of atom symbols  and the number of hydrogens as inputs. The class calculates combinatorially all the possible distributions of these hydrogens. This class has been developed first as a helper class for a structure generator but can be also added to org.openscience.cdk.tools.CDKHydrogenAdder package, if you also think that should be under this CDK package.

The class is shared on GitHub with its test class as a maven project.
HydrogenDistributor: https://github.com/MehmetAzizYirik/HydrogenDistributor

Please let me know if it can be added to this CDK package (or any other CDK packages) or not.

Kind Regards,
Mehmet Aziz Yirik

[Cdk-devel] Nice challenge for SDG ...

[Christoph S. <chr...@un...>]

… put up by Tony Williams in his talk that I am just listening to.

Pinnatoxin A (https://pubchem.ncbi.nlm.nih.gov/compound/178043#section=Names-and-Identifiers)

CC1CC23CCC(=CC2C4C5C(C(CC(O4)(O5)CC6CCCC7(O6)CCC8(O7)C(CCC(O8)CC(=C)CCCC3=NCC1C)(C)O)C)O)C(=O)O

CDK Depict makes the attached picture.



Not bad, but also not too good. 

Cheers, 

Chris

— 
Prof. Dr. Christoph Steinbeck
Analytical Chemistry - Cheminformatics and Chemometrics
Friedrich-Schiller-University Jena, Germany
Phone Secretariat: +49-3641-948171
http://cheminf.uni-jena.de
http://orcid.org/0000-0001-6966-0814

What is man but that lofty spirit - that sense of enterprise.
... Kirk, "I, Mudd," stardate 4513.3..

Re: [Cdk-devel] Potential Bug in Pattern class?

[Egon W. <ego...@gm...>]

On Mon, Aug 13, 2018 at 12:37 AM John Mayfield <joh...@gm...>
wrote:

> I would prefer having something (similar functionality to the
> QueryAtomContainerCreator
> <https://cdk.github.io/cdk/latest/docs/api/org/openscience/cdk/isomorphism/matchers/QueryAtomContainerCreator.html>
> - but with a better name) that takes a molecule and creates the pattern
> given some conditions.
>

Yes, that would be nice... a long standing feature request is a SMARTS
editor, which could be under the hood implemented as something like that...

(it also reminds me of https://smartsview.zbh.uni-hamburg.de/ ...)

Egon

-- 
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: https://www.zotero.org/egonw
ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
ImpactStory: https://impactstory.org/u/egonwillighagen

Re: [Cdk-devel] Potential Bug in Pattern class?

[John M. <joh...@gm...>]

Side note: You shouldn't really be using graph isomorphism to do identity
check, Canonical SMILES/InChI best for that but this is convenient to get
the atom mapping (bijection) between two molecules, you can do this with
canonical IDs (e.g. Canonical SMILES) but you need to "go round the houses"
to get result. Also rather than have different algorithms for sub vs
identity (as this currently is) I would prefer having something (similar
functionality to the QueryAtomContainerCreator
<https://cdk.github.io/cdk/latest/docs/api/org/openscience/cdk/isomorphism/matchers/QueryAtomContainerCreator.html>
- but with a better name) that takes a molecule and creates the pattern
given some conditions. OEChem does it nicely where you parse in the BitSets
of the features you want to match, it then creates a pattern (SMARTS) for
your, i'd probably call it "SmartsFromMol" maybe just to troll RDKit's
"SmartsToMol" (which should really be SmartsToQuery)...

https://docs.eyesopen.com/toolkits/python/oechemtk/OEChemClasses/OESubSearch.html


On Sun, 12 Aug 2018 at 23:26, John Mayfield <joh...@gm...>
wrote:

> No that's not intended, sorry it was updated recently I think. Will fix.
>
> On Sun, 12 Aug 2018 at 10:24, Marwin Segler <mar...@un...>
> wrote:
>
>> Hi everyone,
>>
>> I had quick question about the CDK code.
>>
>> in the isomorphism/Pattern class, there are the `findSubstructure` and
>> `findIdentical` methods.
>>
>> If `Pattern.findSubstructure` calls the underlying
>> `VentoFoggia.findSubstructure`, why does
>>
>> `Pattern.findIdentical` call `VentoFoggia.findSubstructure` as well? Is
>> that intended?
>>
>> Cheers
>> Marwin
>>
>>
>> Code Snippet and link to github below
>>
>>     public static Pattern findSubstructure(IAtomContainer query) {
>>         return VentoFoggia.findSubstructure(query);
>>     }
>>
>>
>>     public static Pattern findIdentical(IAtomContainer query) {
>>         return VentoFoggia.findSubstructure(query);
>>     }
>>
>>
>>
>> https://github.com/cdk/cdk/blob/76625167dedaef688f5bb13278251994cfb3def5/base/isomorphism/src/main/java/org/openscience/cdk/isomorphism/Pattern.java#L180
>
>

Re: [Cdk-devel] Potential Bug in Pattern class?

[John M. <joh...@gm...>]

No that's not intended, sorry it was updated recently I think. Will fix.

On Sun, 12 Aug 2018 at 10:24, Marwin Segler <mar...@un...>
wrote:

> Hi everyone,
>
> I had quick question about the CDK code.
>
> in the isomorphism/Pattern class, there are the `findSubstructure` and
> `findIdentical` methods.
>
> If `Pattern.findSubstructure` calls the underlying
> `VentoFoggia.findSubstructure`, why does
>
> `Pattern.findIdentical` call `VentoFoggia.findSubstructure` as well? Is
> that intended?
>
> Cheers
> Marwin
>
>
> Code Snippet and link to github below
>
>     public static Pattern findSubstructure(IAtomContainer query) {
>         return VentoFoggia.findSubstructure(query);
>     }
>
>
>     public static Pattern findIdentical(IAtomContainer query) {
>         return VentoFoggia.findSubstructure(query);
>     }
>
>
>
> https://github.com/cdk/cdk/blob/76625167dedaef688f5bb13278251994cfb3def5/base/isomorphism/src/main/java/org/openscience/cdk/isomorphism/Pattern.java#L180

[Cdk-devel] Potential Bug in Pattern class?

[Marwin S. <mar...@un...>]

Hi everyone,

I had quick question about the CDK code.

in the isomorphism/Pattern class, there are the `findSubstructure` and `findIdentical` methods.

If `Pattern.findSubstructure` calls the underlying `VentoFoggia.findSubstructure`, why does

`Pattern.findIdentical` call `VentoFoggia.findSubstructure` as well? Is that intended?

Cheers
Marwin


Code Snippet and link to github below

    public static Pattern findSubstructure(IAtomContainer query) {
        return VentoFoggia.findSubstructure(query);
    }


    public static Pattern findIdentical(IAtomContainer query) {
        return VentoFoggia.findSubstructure(query);
    }


https://github.com/cdk/cdk/blob/76625167dedaef688f5bb13278251994cfb3def5/base/isomorphism/src/main/java/org/openscience/cdk/isomorphism/Pattern.java#L180

Re: [Cdk-devel] TemplateHandler3D issue

[John M. <joh...@gm...>]

Yes looks reasonable, will update. In general the 3D code is in a very
sorry state to the point where I really would not recomend using it, but if
it works for you feel free.

John

On 12 March 2018 at 09:58, Ivaylo Kirov via Cdk-devel <
cdk...@li...> wrote:

> Greetings,
>
> I found an issue with the templateHandler logic and I will appreciate if
> you could support me on that one.
>
> File: TemplateHandler3D.cs
> Line: 227
>
> The warning pop-ups every time on generating 3D ring coordinates which
> seems a little bit weird.
> So I was thinking is it possible to be a bug?
>
> Original code:
>
>             if (best != null)
>             {
>                 AssignCoords(best, bestMap);
>             }
>             else if (secondBest != null)
>             {
>                 AssignCoords(secondBest, secondBestMap);
>             }
>
>             Console.Error.WriteLine("WARNING: Maybe RingTemplateError!");
>
> Shouldn't it be something like that:
>
>             if (best != null)
>             {
>                 AssignCoords(best, bestMap);
>             }
>             else if (secondBest != null)
>             {
>                 AssignCoords(secondBest, secondBestMap);
>             }
>             else
>             {
>             Console.Error.WriteLine("WARNING: Maybe RingTemplateError!");
>             }
>
> Or even that one (showing warning if there is no primal best make sense
> too):
>
>             if (best != null)
>             {
>                 AssignCoords(best, bestMap);
>             }
>             else if (secondBest != null)
>             {
>                 AssignCoords(secondBest, secondBestMap);
>
>                 Console.Error.WriteLine("WARNING: Maybe
> RingTemplateError!");
>             }
>
>
>
> Thank you in advance
>
> Best regards
>
> Ivaylo
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>

[Cdk-devel] TemplateHandler3D issue

[Ivaylo K. <ivo...@ya...>]

Greetings,
I found an issue with the templateHandler logic and I will appreciate if you could support me on that one.
File: TemplateHandler3D.csLine: 227
The warning pop-ups every time on generating 3D ring coordinates which seems a little bit weird.So I was thinking is it possible to be a bug?
Original code:
            if (best != null)            {                AssignCoords(best, bestMap);            }            else if (secondBest != null)            {                AssignCoords(secondBest, secondBestMap);            }
            Console.Error.WriteLine("WARNING: Maybe RingTemplateError!");
Shouldn't it be something like that:
            if (best != null)            {                AssignCoords(best, bestMap);            }            else if (secondBest != null)            {                AssignCoords(secondBest, secondBestMap);            }            else
            {            Console.Error.WriteLine("WARNING: Maybe RingTemplateError!");            }
Or even that one (showing warning if there is no primal best make sense too):
            if (best != null)            {                AssignCoords(best, bestMap);            }            else if (secondBest != null)            {                AssignCoords(secondBest, secondBestMap);
                Console.Error.WriteLine("WARNING: Maybe RingTemplateError!");
            }            


Thank you in advance
Best regards
Ivaylo

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Nina J. <jel...@gm...>]

On 11 March 2018 at 18:41, Egon Willighagen <ego...@gm...>
wrote:

>
>
> On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
> chr...@un...> wrote:
>
>> With regard to your statement below: Is there a chance to disentangle the
>> actual CDK nodes from JCP and AMBIT?
>> For JCP, there is the excellent Marvin Sketch node as a temporary
>> replacement.
>>
>
> That sounds like a reasonable back up...
>
>
>> Not sure if the AMBIT developers can help migrate to 2.x or move to a
>> dedicated repository?
>>
>
> I think we can get AMBIT to work... I will coordinate this with Nina.
>

Yes, we already have a branch with CDK2 dependency
<http://svn.code.sf.net/p/ambit/code/branches/ambit-cdk-2/> , there are
some failing tests, mostly related to stereochemistry.

Regards,
Nina


>
> Egon
>
>
> --
> E.L. Willighagen
> Department of Bioinformatics - BiGCaT
> Maastricht University (http://www.bigcat.unimaas.nl/)
> Homepage: http://egonw.github.com/
> LinkedIn: http://se.linkedin.com/in/egonw
> Blog: http://chem-bla-ics.blogspot.com/
> PubList: https://www.zotero.org/egonw
> ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
> ImpactStory: https://impactstory.org/u/egonwillighagen
>

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Egon W. <ego...@gm...>]

Do we have a JChemPaint code base based on CDK 2?

Egon

On Sun, Mar 11, 2018 at 9:36 PM, John Mayfield <joh...@gm...>
wrote:

> CDK 1.5.13 -> 2.1 should be relatively painless.
>
> On 11 March 2018 at 16:41, Egon Willighagen <ego...@gm...>
> wrote:
>
>>
>>
>> On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
>> chr...@un...> wrote:
>>
>>> With regard to your statement below: Is there a chance to disentangle
>>> the actual CDK nodes from JCP and AMBIT?
>>> For JCP, there is the excellent Marvin Sketch node as a temporary
>>> replacement.
>>>
>>
>> That sounds like a reasonable back up...
>>
>>
>>> Not sure if the AMBIT developers can help migrate to 2.x or move to a
>>> dedicated repository?
>>>
>>
>> I think we can get AMBIT to work... I will coordinate this with Nina.
>>
>> Egon
>>
>>
>> --
>> E.L. Willighagen
>> Department of Bioinformatics - BiGCaT
>> Maastricht University (http://www.bigcat.unimaas.nl/)
>> Homepage: http://egonw.github.com/
>> LinkedIn: http://se.linkedin.com/in/egonw
>> Blog: http://chem-bla-ics.blogspot.com/
>> PubList: https://www.zotero.org/egonw
>> ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
>> ImpactStory: https://impactstory.org/u/egonwillighagen
>>
>> ------------------------------------------------------------
>> ------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>> _______________________________________________
>> Cdk-devel mailing list
>> Cdk...@li...
>> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>>
>>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>


-- 
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: https://www.zotero.org/egonw
ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
ImpactStory: https://impactstory.org/u/egonwillighagen

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[John M. <joh...@gm...>]

CDK 1.5.13 -> 2.1 should be relatively painless.

On 11 March 2018 at 16:41, Egon Willighagen <ego...@gm...>
wrote:

>
>
> On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
> chr...@un...> wrote:
>
>> With regard to your statement below: Is there a chance to disentangle the
>> actual CDK nodes from JCP and AMBIT?
>> For JCP, there is the excellent Marvin Sketch node as a temporary
>> replacement.
>>
>
> That sounds like a reasonable back up...
>
>
>> Not sure if the AMBIT developers can help migrate to 2.x or move to a
>> dedicated repository?
>>
>
> I think we can get AMBIT to work... I will coordinate this with Nina.
>
> Egon
>
>
> --
> E.L. Willighagen
> Department of Bioinformatics - BiGCaT
> Maastricht University (http://www.bigcat.unimaas.nl/)
> Homepage: http://egonw.github.com/
> LinkedIn: http://se.linkedin.com/in/egonw
> Blog: http://chem-bla-ics.blogspot.com/
> PubList: https://www.zotero.org/egonw
> ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
> ImpactStory: https://impactstory.org/u/egonwillighagen
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Egon W. <ego...@gm...>]

On Sun, Mar 11, 2018 at 5:03 PM, Christoph Steinbeck <
chr...@un...> wrote:

> With regard to your statement below: Is there a chance to disentangle the
> actual CDK nodes from JCP and AMBIT?
> For JCP, there is the excellent Marvin Sketch node as a temporary
> replacement.
>

That sounds like a reasonable back up...


> Not sure if the AMBIT developers can help migrate to 2.x or move to a
> dedicated repository?
>

I think we can get AMBIT to work... I will coordinate this with Nina.

Egon


-- 
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: https://www.zotero.org/egonw
ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
ImpactStory: https://impactstory.org/u/egonwillighagen

Re: [Cdk-devel] KNIME-CDK without failing unit tests

[Christoph S. <chr...@un...>]

Great news to see the migration of KNIME-CDK to 2.x move forward. 

With regard to your statement below: Is there a chance to disentangle the actual CDK nodes from JCP and AMBIT?
For JCP, there is the excellent Marvin Sketch node as a temporary replacement. 

Not sure if the AMBIT developers can help migrate to 2.x or move to a dedicated repository?

Kind regards, 

Chris
 
— 
Prof. Dr. Christoph Steinbeck
Analytical Chemistry - Cheminformatics and Chemometrics
Friedrich-Schiller-University Jena, Germany
Phone Secretariat: +49-3641-948171
http://cheminf.uni-jena.de
http://orcid.org/0000-0001-6966-0814

What is man but that lofty spirit - that sense of enterprise.
... Kirk, "I, Mudd," stardate 4513.3..

> On 11 Mar 02018, at 16:11, Egon Willighagen <ego...@gm...> wrote:
> 
> 
> Hi all,
> 
> last autumn I participated in the KNIME workshop at the German Cheminformatics Conference, met up with Thorsten (in a long time) and we agreed to move the CDK nodes to its own repository, which since lives under:
> 
> https://github.com/cdk/nodes4knime
> 
> Now that no tests fail anymore, we can start thinking about updates. But since this repository actually has CDK nodes, but also JChemPaint and AMBIT nodes, the upgrade path may be a bit tricky...
> 
> Egon
> 
> -- 
> E.L. Willighagen
> Department of Bioinformatics - BiGCaT
> Maastricht University (http://www.bigcat.unimaas.nl/)
> Homepage: http://egonw.github.com/
> LinkedIn: http://se.linkedin.com/in/egonw
> Blog: http://chem-bla-ics.blogspot.com/
> PubList: https://www.zotero.org/egonw
> ORCID: 0000-0001-7542-0286
> ImpactStory: https://impactstory.org/u/egonwillighagen
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot_______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel

[Cdk-devel] KNIME-CDK without failing unit tests

[Egon W. <ego...@gm...>]

Hi all,

last autumn I participated in the KNIME workshop at the German
Cheminformatics Conference, met up with Thorsten (in a long time) and we
agreed to move the CDK nodes to its own repository, which since lives under:

https://github.com/cdk/nodes4knime

Now that no tests fail anymore, we can start thinking about updates. But
since this repository actually has CDK nodes, but also JChemPaint and AMBIT
nodes, the upgrade path may be a bit tricky...

Egon

-- 
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: https://www.zotero.org/egonw
ORCID: 0000-0001-7542-0286 <http://orcid.org/0000-0001-7542-0286>
ImpactStory: https://impactstory.org/u/egonwillighagen

Re: [Cdk-devel] CDK unique smiles flavour fails on some cyclobutadiens

[John M. <joh...@gm...>]

Yes, by design and matches the InChI behaviour.

CC1=C(C)C=C1
InChI=1S/C6H8/c1-5-3-4-6(5)2/h3-4H,1-2H3
InChIKey=QNZUSULCBOVZRK-UHFFFAOYSA-N

CC1=CC=C1C
InChI=1S/C6H8/c1-5-3-4-6(5)2/h3-4H,1-2H3
InChIKey=QNZUSULCBOVZRK-UHFFFAOYSA-N

Instead of looking at aromaticity (a whole can of worms) conjugated double
bonds in rings are allowed to move irrespective of their aromatic
status (conjugate
equivalence). Since SMILES/InChI are primarily meant for information
retrieval it's useful to get one from the other. If you need further
refinement you can do a post match check. We *could* add an option that
only allows bonds marked as aromatic to migrate. A better example is
something like this:


If you want to force differences, define the stereochemistry on the double
bonds. This says "don't move this double bond", again matching the InChI.

- John

On 8 March 2018 at 10:37, Christoph Steinbeck <
chr...@un...> wrote:

> I was looking at whether RandomGenerator covers all of constitutional
> isomer space that it is supposed to do (it seems to do) and used CDK unique
> smiles as a hash. In this test code, RandomGenerator found only 4308 of the
> 4315 compounds that Molgen (http://www.molgen.de/literature.html)
> predicted.
>
> Turned out that the reason was that the CDK unique smiles implementation
> generates identical ‘unique’ smiles for non-identical molecules. They are
> all mesomeric cyclobutadiens. See attached graphics and SDF file.
>
> Both RDKIT and Openbabel generate the right number 4315 non-identical
> unique smiles.
>
> Kind regards,
>
> Chris
>
> —
> Prof. Dr. Christoph Steinbeck
> Analytical Chemistry - Cheminformatics and Chemometrics
> Friedrich-Schiller-University Jena, Germany
> Phone Secretariat: +49-3641-948171
> http://cheminf.uni-jena.de
> http://orcid.org/0000-0001-6966-0814
>
> What is man but that lofty spirit - that sense of enterprise.
> ... Kirk, "I, Mudd," stardate 4513.3..
>
>
>
>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>

[Cdk-devel] CDK unique smiles flavour fails on some cyclobutadiens

[Christoph S. <chr...@un...>]

I was looking at whether RandomGenerator covers all of constitutional isomer space that it is supposed to do (it seems to do) and used CDK unique smiles as a hash. In this test code, RandomGenerator found only 4308 of the 4315 compounds that Molgen (http://www.molgen.de/literature.html) predicted. 

Turned out that the reason was that the CDK unique smiles implementation generates identical ‘unique’ smiles for non-identical molecules. They are all mesomeric cyclobutadiens. See attached graphics and SDF file.

Both RDKIT and Openbabel generate the right number 4315 non-identical unique smiles. 

Kind regards, 

Chris

— 
Prof. Dr. Christoph Steinbeck
Analytical Chemistry - Cheminformatics and Chemometrics
Friedrich-Schiller-University Jena, Germany
Phone Secretariat: +49-3641-948171
http://cheminf.uni-jena.de
http://orcid.org/0000-0001-6966-0814

What is man but that lofty spirit - that sense of enterprise.
... Kirk, "I, Mudd," stardate 4513.3..

Re: [Cdk-devel] Adding a descriptor to the CDK

[Nina J. <jel...@gm...>]

Hi Andy, John,

One creates new molecular descriptors compatible with CDK by implementing
IMolecularDescriptor
<http://cdk.github.io/cdk/2.1/docs/api/index.html?org/openscience/cdk/qsar/IMolecularDescriptor.html>
interface,
as already mentioned by John.  You could look for other classes of
molecular descriptors for examples, it basically comes down to implementing
few methods defined by the interface.

It is of course possible to implement descriptors outside of CDK (we have
done it a lot in Ambit) , but it's not necessarily convenient for all users.

As for where the new code should go, I have no strong preference, but
descriptors have been contributed before (yes, some are a headache) ,
updated, maintained or not, and so on.
I also am not quite sure why the descriptor by Andy should be treated in a
different way than already contributed descriptors by others, and how does
one decide that something goes into core CDK and what goes into Contrib.

In any case, would be best to agree on some procedure and put a guidance
online, if it will be different from just forking the CDK and sending a
pull request with the new descriptor.

Best regards,
Nina


On 18 January 2018 at 22:08, John Mayfield <joh...@gm...>
wrote:

> Hi Andrew,
>
> It depends what you ultimate end game is... (i.e. why?) I would be very
> hesitant to add complex *models* as *descriptors* which is what I think
> you're suggesting. There are some already (e.g. IonisationPotential IP,
> ALogP) but they're nothing but a headache, not very accurate, and buggy
> (for example the IP changes based on the storage order of atoms in a
> bond!). I tend to think the *descriptors* should be "primitive"
> properties for example nitrogen/oxygen balance, linear paths, ring sizes, h
> donor/acceptor, radial fingerprints, etc, etc. Users can then build their
> *models* down stream based on these values (we even have the *cdk-model *module
> for simple Baysian models).
>
> You can implement and distribute *models *as *descritpros* outside of CDK
> in a portable and pluggable manner (extending IMolecularDescriptor
> <http://cdk.github.io/cdk/2.1/docs/api/index.html?org/openscience/cdk/qsar/IMolecularDescriptor.html>)
> and if they prove to be accurate and useful then they could be merged into
> the main CDK. We've recently be discussing the idea of Contrib location and
> this use would fit nicely. Perhaps others disagree?
>
> Best Wishes,
> John
>
> P.S. For your example the SubstructureFingerprint
> <http://cdk.github.io/cdk/2.1/docs/api/index.html?org/openscience/cdk/fingerprint/SubstructureFingerprinter.html>
>  can generate a binary descriptor based on pre-define functional groups
> (or you can define your own).
>
> On 18 January 2018 at 18:35, Andrew Lang <asi...@gm...> wrote:
>
>> Hi Guys:
>>
>> My end goal is to add new descriptors to the CDK.
>>
>> I can:
>>
>> Generate a list of fragments using: http://vpsolovev.ru/programs/smf/
>>
>> and
>>
>> Use R to build a linear model
>>
>> What I'd like to know is:
>>
>> 1. Are these fragments the way to go (prefered method) when one wants to
>> develop a new descriptor for the CDK.
>>
>> 2. How does one add a descriptor to the CDK given my programming skills
>> are non-expert level.
>>
>> Thanks,
>>
>> Andy
>>
>> Andrew Lang
>> Professor and Chair, Mathematics
>> Oral Roberts University
>>
>>
>>
>> ------------------------------------------------------------
>> ------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>> _______________________________________________
>> Cdk-devel mailing list
>> Cdk...@li...
>> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>>
>>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>

Re: [Cdk-devel] Adding a descriptor to the CDK

[John M. <joh...@gm...>]

Hi Andrew,

It depends what you ultimate end game is... (i.e. why?) I would be very
hesitant to add complex *models* as *descriptors* which is what I think
you're suggesting. There are some already (e.g. IonisationPotential IP,
ALogP) but they're nothing but a headache, not very accurate, and buggy
(for example the IP changes based on the storage order of atoms in a
bond!). I tend to think the *descriptors* should be "primitive" properties
for example nitrogen/oxygen balance, linear paths, ring sizes, h
donor/acceptor, radial fingerprints, etc, etc. Users can then build their
*models* down stream based on these values (we even have the *cdk-model *module
for simple Baysian models).

You can implement and distribute *models *as *descritpros* outside of CDK
in a portable and pluggable manner (extending IMolecularDescriptor
<http://cdk.github.io/cdk/2.1/docs/api/index.html?org/openscience/cdk/qsar/IMolecularDescriptor.html>)
and if they prove to be accurate and useful then they could be merged into
the main CDK. We've recently be discussing the idea of Contrib location and
this use would fit nicely. Perhaps others disagree?

Best Wishes,
John

P.S. For your example the SubstructureFingerprint
<http://cdk.github.io/cdk/2.1/docs/api/index.html?org/openscience/cdk/fingerprint/SubstructureFingerprinter.html>
can
generate a binary descriptor based on pre-define functional groups (or you
can define your own).

On 18 January 2018 at 18:35, Andrew Lang <asi...@gm...> wrote:

> Hi Guys:
>
> My end goal is to add new descriptors to the CDK.
>
> I can:
>
> Generate a list of fragments using: http://vpsolovev.ru/programs/smf/
>
> and
>
> Use R to build a linear model
>
> What I'd like to know is:
>
> 1. Are these fragments the way to go (prefered method) when one wants to
> develop a new descriptor for the CDK.
>
> 2. How does one add a descriptor to the CDK given my programming skills
> are non-expert level.
>
> Thanks,
>
> Andy
>
> Andrew Lang
> Professor and Chair, Mathematics
> Oral Roberts University
>
>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Cdk-devel mailing list
> Cdk...@li...
> https://lists.sourceforge.net/lists/listinfo/cdk-devel
>
>

[Cdk-devel] Adding a descriptor to the CDK

[Andrew L. <asi...@gm...>]

Hi Guys:

My end goal is to add new descriptors to the CDK.

I can:

Generate a list of fragments using: http://vpsolovev.ru/programs/smf/

and

Use R to build a linear model

What I'd like to know is:

1. Are these fragments the way to go (prefered method) when one wants to
develop a new descriptor for the CDK.

2. How does one add a descriptor to the CDK given my programming skills are
non-expert level.

Thanks,

Andy

Andrew Lang
Professor and Chair, Mathematics
Oral Roberts University

[Cdk-devel] Request for comment: AtomContainer2

[John M. <joh...@gm...>]

Hi All,

I've been working on a patch since March that greatly enhances the CDK
structure manipulation both in terms of usability and performance.

It is essentially backwards compatible with the current API (barring some
minor caveats) but to minimise issues I was planning a gradual introduction
where by initially it is turned on and off using an system/environment flag
and removing the flag and current implementation as a 3.0 release.

In summary you can now do this:

IAtomContainer mol = ...;
> for (IAtom atom : mol.atoms()) {
>   int aidx = atom.getIndex(); // O(1) constant time!
>   int deg = atom.getBondCount(); // O(1) constant time!
>   // atoms know about their bonds O(|degree|)
>   for (IBond bond : atom.bonds()) {
>     int bidx = bond.getIndex(); // O(1) constant time!
>   }
> }


I believe it should be relatively pain free to migrate however only testing
will tell. This is where we need your help! Please review the patch and
once applied please test you code with the new implementation (see info
below).

Further details/benchmarks: https://github.com/cdk/cdk/wiki/AtomContainer2
Patch: https://github.com/cdk/cdk/pull/368

Many Thanks,
John