From: Ola S. <ola...@fa...> - 2007-12-21 11:36:59
|
[Cross-posted to cdk-devel and bioclipse-devel] Hi, I am working on the QSAR plugin for Bioclipse2 (see my post: <http:// bioclipse.blogspot.com/2007/12/towards-standards-for-qsar-data- setup.html>) and have some questions. 1) Is it possible to publish molecules to a public repository and get REST ID's to them (something like below)? http://www.mol.repo.org/molecules/cml/id?abc234 http://www.mol.repo.org/molecules/mol2/id?abc234 Where can I upload/download them in that case? 2) Does it exist (or could we establish) fixed URL's to be used as REST ID's for descriptors? Isn't there a BO-ontology for descriptors online? I also want to be able to get that RestID (=URL) via the CDK descriptor API: myCDKDescritptor.getURL() would yield something like: "http:// www.cdk.sf.net/descriptors/xlogp:implementation1" The goal is to just have the qsar.xml (see blog post in top) and be able to download all molecules from a public repository, and automatically calculate the descriptors from within Bioclipse. Cheers, .../Ola |
From: Rajarshi G. <rg...@in...> - 2007-12-21 13:42:33
|
On Dec 21, 2007, at 6:38 AM, Ola Spjuth wrote: > 1) > Is it possible to publish molecules to a public repository and get > REST ID's to them (something like below)? > http://www.mol.repo.org/molecules/cml/id?abc234 > http://www.mol.repo.org/molecules/mol2/id?abc234 > > Where can I upload/download them in that case? It can be done. You might look at Chemspider - I think they offer access to their DB in a REST format. If it's useful, we provide REST access to our PubChem mirror - access would be via compound ID (CID) > 2) > Does it exist (or could we establish) fixed URL's to be used as REST > ID's for descriptors? Isn't there a BO-ontology for descriptors > online? I also want to be able to get that RestID (=URL) via the CDK > descriptor API: > > myCDKDescritptor.getURL() would yield something like: "http:// > www.cdk.sf.net/descriptors/xlogp:implementation1 Good to see someone wants this :) We recently put in a grant application for a superset of this problem - part of the grant described an approach for automated, distributed descriptor calculations. First, we have a CDK desc web service (needs some work to get it in sync with the current CDK). It is easy to put a REST interface on top of it. The CDK desc ontology does include specification references that includes descriptor name and provider. So all that is left is to have a resolver at a constant location, say the URL that you provided. The resolver would have to know about the location of various implementers and would simply rewrite the incoming URL to that of the provider URL. There would be some design issues: is the URL designed for a single molecule? Or multiple molecules? Will it take SMILES? SDF? Both? In fact the resolver, could simply be an Apache rewrite rule :) Would involve a hairy regex though! But yes, it could certainly be established. ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- Alcohol, an alternative to your self - 'Alcohol' by the Bare Naked Ladies |
From: Egon W. <ego...@gm...> - 2008-01-02 21:49:13
|
On Dec 21, 2007 12:38 PM, Ola Spjuth <ola...@fa...> wrote: > 1) > Is it possible to publish molecules to a public repository and get > REST ID's to them (something like below)? > http://www.mol.repo.org/molecules/cml/id?abc234 > http://www.mol.repo.org/molecules/mol2/id?abc234 PubChem? > Where can I upload/download them in that case? http://pubchem.ncbi.nlm.nih.gov/deposit/deposit.cgi > 2) > Does it exist (or could we establish) fixed URL's to be used as REST > ID's for descriptors? Isn't there a BO-ontology for descriptors > online? Yes: http://qsar.sourceforge.net/dicts/qsar-descriptors/index.xhtml The source is available in OWL format from: http://qsar.svn.sourceforge.net/viewvc/qsar/trunk/qsar-dicts/descriptor-algorithms.owl?revision=202&view=markup Being RDF, each 'descriptor' de facto has an URL... For example, the URL for the atomDegree descriptor is: http://www.blueobelisk.org/ontologies/chemoinformatics-algorithms/#atomDegree > I also want to be able to get that RestID (=URL) via the CDK > descriptor API: > > myCDKDescritptor.getURL() would yield something like: "http:// > www.cdk.sf.net/descriptors/xlogp:implementation1" Check out the getSpecification() method... that's provides what you are looking for. > The goal is to just have the qsar.xml (see blog post in top) and be > able to download all molecules from a public repository, InChIs... It's easy to make Bioclipse aware of where to download/get what for a certain InChI... > and automatically calculate the descriptors from within Bioclipse. Egon -- ---- http://chem-bla-ics.blogspot.com/ |
From: Rajarshi G. <rg...@in...> - 2008-01-07 05:20:34
|
On Dec 21, 2007, at 8:42 AM, Rajarshi Guha wrote: >> 2) >> Does it exist (or could we establish) fixed URL's to be used as REST >> ID's for descriptors? Isn't there a BO-ontology for descriptors >> online? I also want to be able to get that RestID (=URL) via the CDK >> descriptor API: >> >> myCDKDescritptor.getURL() would yield something like: "http:// >> www.cdk.sf.net/descriptors/xlogp:implementation1 > > First, we have a CDK desc web service (needs some work to get it in > sync with the current CDK). It is easy to put a REST interface on top > of it. The CDK desc ontology does include specification references > that includes descriptor name and provider. A web service is now available that calculates descriptors. The WSDL is located at http://rguha.ath.cx:8080/cdkws/services/Descriptors? wsdl and API docs are at http://cheminfo.informatics.indiana.edu/ ~rguha/ws/cdkws/api/ The relevant functions are getAvailableDescriptorNames(String) which allows you to get back the class names of all or a subset (topological, geometrical etc) of descriptors. These names can then be used with evaluateDescriptors(String[], String). This method accepts a single SMILES or a single molecule SDF (in the form of a string). The return value is an XML document containing the descriptor value(s). Ideally, I'd make this CML, but for now it's very trivial XML. > So all that is left is to have a resolver at a constant location, say > the URL that you provided. The resolver would have to know about the > location of various implementers and would simply rewrite the > incoming URL to that of the provider URL. Right now there is no resolver, but a REST interface is available for descriptor calculation. > There would be some design issues: is the URL designed for a single > molecule? Or multiple molecules? Will it take SMILES? SDF? Both? The entry URI for the REST interface is for a single molecule (which I think is RESTful) and only accepts SMILES (though an SDF string should work) and will only evaluate topological and constitutional descriptors (I plan on updating this to the full set of descriptors, but requires some thinking on how to handle errors such as when a 3D descriptor is attempted for a SMILES string) http://rguha.ath.cx/~rguha/cicc/desc/descriptors/CCOCC(O)=O This will return an XML document containing a series of descriptor- ref elements, whose href attribute is a URL that will return the descriptor values in XML form. So for example one of the elements from the above document would be <descriptor-ref xlink:href="http://rguha.ath.cx/~rguha/cicc/desc/ descriptors/ org.openscience.cdk.qsar.descriptors.molecular.XLogPDescriptor/CCOCC (O)=O"/> Visiting the URL in the href attribute would return <?xml version="1.0"?> <DescriptorList> <Descriptor parent="XLogPDescriptor" name="XLogP" value="-0.744"/> </DescriptorList> I should point out that I'm new to REST style programming, so I may have interpreted stuff wrong - feel free to point out mistakes. ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- A red sign on the door of a physics professor: 'If this sign is blue, you're going too fast.' |
From: Egon W. <ego...@gm...> - 2008-01-07 06:54:53
|
So, Ola you still like to see these integrated into the Bioclipse-QSAR framework? Egon On Jan 7, 2008 6:20 AM, Rajarshi Guha <rg...@in...> wrote: > > On Dec 21, 2007, at 8:42 AM, Rajarshi Guha wrote: > > >> 2) > >> Does it exist (or could we establish) fixed URL's to be used as REST > >> ID's for descriptors? Isn't there a BO-ontology for descriptors > >> online? I also want to be able to get that RestID (=URL) via the CDK > >> descriptor API: > >> > >> myCDKDescritptor.getURL() would yield something like: "http:// > >> www.cdk.sf.net/descriptors/xlogp:implementation1 > > > > First, we have a CDK desc web service (needs some work to get it in > > sync with the current CDK). It is easy to put a REST interface on top > > of it. The CDK desc ontology does include specification references > > that includes descriptor name and provider. > > A web service is now available that calculates descriptors. The WSDL > is located at http://rguha.ath.cx:8080/cdkws/services/Descriptors? > wsdl and API docs are at http://cheminfo.informatics.indiana.edu/ > ~rguha/ws/cdkws/api/ > > The relevant functions are getAvailableDescriptorNames(String) which > allows you to get back the class names of all or a subset > (topological, geometrical etc) of descriptors. > > These names can then be used with evaluateDescriptors(String[], > String). This method accepts a single SMILES or a single molecule SDF > (in the form of a string). The return value is an XML document > containing the descriptor value(s). Ideally, I'd make this CML, but > for now it's very trivial XML. > > > So all that is left is to have a resolver at a constant location, say > > the URL that you provided. The resolver would have to know about the > > location of various implementers and would simply rewrite the > > incoming URL to that of the provider URL. > > Right now there is no resolver, but a REST interface is available for > descriptor calculation. > > > There would be some design issues: is the URL designed for a single > > molecule? Or multiple molecules? Will it take SMILES? SDF? Both? > > The entry URI for the REST interface is for a single molecule (which > I think is RESTful) and only accepts SMILES (though an SDF string > should work) and will only evaluate topological and constitutional > descriptors (I plan on updating this to the full set of descriptors, > but requires some thinking on how to handle errors such as when a 3D > descriptor is attempted for a SMILES string) > > http://rguha.ath.cx/~rguha/cicc/desc/descriptors/CCOCC(O)=O > > This will return an XML document containing a series of descriptor- > ref elements, whose href attribute is a URL that will return the > descriptor values in XML form. > > So for example one of the elements from the above document would be > > <descriptor-ref xlink:href="http://rguha.ath.cx/~rguha/cicc/desc/ > descriptors/ > org.openscience.cdk.qsar.descriptors.molecular.XLogPDescriptor/CCOCC > (O)=O"/> > > Visiting the URL in the href attribute would return > > <?xml version="1.0"?> > <DescriptorList> > <Descriptor parent="XLogPDescriptor" name="XLogP" value="-0.744"/> > </DescriptorList> > > I should point out that I'm new to REST style programming, so I may > have interpreted stuff wrong - feel free to point out mistakes. -- ---- http://chem-bla-ics.blogspot.com/ |
From: Ola S. <ola...@fa...> - 2008-01-07 14:58:20
|
Egon: Yes. I will need REST URL's that define the descriptors, will use BO ontology for now. Additionally it would be great to also provide information about an implementation (like use local CDK or use remote invocation via Web services) but that is not my primary objective. If we share the same REST ID's then Rajarshi's Web services could be plugged into Bioclipse and be available as a separate descriptor provider (in addition to CDK and JOELib). Should be easy to set up. Link: http://wiki.bioclipse.net/index.php? title=QSAR_Project_with_automated_builds /Ola On Jan 7, 2008, at 07:54 , Egon Willighagen wrote: > So, Ola you still like to see these integrated into the Bioclipse-QSAR > framework? > > Egon > > On Jan 7, 2008 6:20 AM, Rajarshi Guha <rg...@in...> wrote: >> >> On Dec 21, 2007, at 8:42 AM, Rajarshi Guha wrote: >> >>>> 2) >>>> Does it exist (or could we establish) fixed URL's to be used as >>>> REST >>>> ID's for descriptors? Isn't there a BO-ontology for descriptors >>>> online? I also want to be able to get that RestID (=URL) via the >>>> CDK >>>> descriptor API: >>>> >>>> myCDKDescritptor.getURL() would yield something >>>> like: "http:// >>>> www.cdk.sf.net/descriptors/xlogp:implementation1 >>> >>> First, we have a CDK desc web service (needs some work to get it in >>> sync with the current CDK). It is easy to put a REST interface on >>> top >>> of it. The CDK desc ontology does include specification references >>> that includes descriptor name and provider. >> >> A web service is now available that calculates descriptors. The WSDL >> is located at http://rguha.ath.cx:8080/cdkws/services/Descriptors? >> wsdl and API docs are at http://cheminfo.informatics.indiana.edu/ >> ~rguha/ws/cdkws/api/ >> >> The relevant functions are getAvailableDescriptorNames(String) which >> allows you to get back the class names of all or a subset >> (topological, geometrical etc) of descriptors. >> >> These names can then be used with evaluateDescriptors(String[], >> String). This method accepts a single SMILES or a single molecule SDF >> (in the form of a string). The return value is an XML document >> containing the descriptor value(s). Ideally, I'd make this CML, but >> for now it's very trivial XML. >> >>> So all that is left is to have a resolver at a constant location, >>> say >>> the URL that you provided. The resolver would have to know about the >>> location of various implementers and would simply rewrite the >>> incoming URL to that of the provider URL. >> >> Right now there is no resolver, but a REST interface is available for >> descriptor calculation. >> >>> There would be some design issues: is the URL designed for a single >>> molecule? Or multiple molecules? Will it take SMILES? SDF? Both? >> >> The entry URI for the REST interface is for a single molecule (which >> I think is RESTful) and only accepts SMILES (though an SDF string >> should work) and will only evaluate topological and constitutional >> descriptors (I plan on updating this to the full set of descriptors, >> but requires some thinking on how to handle errors such as when a 3D >> descriptor is attempted for a SMILES string) >> >> http://rguha.ath.cx/~rguha/cicc/desc/descriptors/CCOCC(O)=O >> >> This will return an XML document containing a series of descriptor- >> ref elements, whose href attribute is a URL that will return the >> descriptor values in XML form. >> >> So for example one of the elements from the above document would be >> >> <descriptor-ref xlink:href="http://rguha.ath.cx/~rguha/cicc/desc/ >> descriptors/ >> org.openscience.cdk.qsar.descriptors.molecular.XLogPDescriptor/CCOCC >> (O)=O"/> >> >> Visiting the URL in the href attribute would return >> >> <?xml version="1.0"?> >> <DescriptorList> >> <Descriptor parent="XLogPDescriptor" name="XLogP" value="-0.744"/> >> </DescriptorList> >> >> I should point out that I'm new to REST style programming, so I may >> have interpreted stuff wrong - feel free to point out mistakes. > > > > -- > ---- > http://chem-bla-ics.blogspot.com/ > > ---------------------------------------------------------------------- > --- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2005. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Cdk-devel mailing list > Cdk...@li... > https://lists.sourceforge.net/lists/listinfo/cdk-devel |
From: Ola S. <ola...@fa...> - 2008-01-07 14:58:40
|
On Jan 2, 2008, at 22:49 , Egon Willighagen wrote: > On Dec 21, 2007 12:38 PM, Ola Spjuth <ola...@fa...> wrote: >> 1) >> Is it possible to publish molecules to a public repository and get >> REST ID's to them (something like below)? >> http://www.mol.repo.org/molecules/cml/id?abc234 >> http://www.mol.repo.org/molecules/mol2/id?abc234 > > PubChem? I don't think it's currently possible in Pubchem. Rajarshi: If you could add REST IDs to your Pubchem via CID that would be great! Would it be possible to go via InChi as well? And what would it return, an SDF or a SMILES string or what? > >> Where can I upload/download them in that case? > > http://pubchem.ncbi.nlm.nih.gov/deposit/deposit.cgi > >> 2) >> Does it exist (or could we establish) fixed URL's to be used as REST >> ID's for descriptors? Isn't there a BO-ontology for descriptors >> online? > > Yes: > > http://qsar.sourceforge.net/dicts/qsar-descriptors/index.xhtml > > The source is available in OWL format from: > > http://qsar.svn.sourceforge.net/viewvc/qsar/trunk/qsar-dicts/ > descriptor-algorithms.owl?revision=202&view=markup > > Being RDF, each 'descriptor' de facto has an URL... > > For example, the URL for the atomDegree descriptor is: > > http://www.blueobelisk.org/ontologies/chemoinformatics-algorithms/ > #atomDegree Right. > >> I also want to be able to get that RestID (=URL) via the CDK >> descriptor API: >> >> myCDKDescritptor.getURL() would yield something >> like: "http:// >> www.cdk.sf.net/descriptors/xlogp:implementation1" > > Check out the getSpecification() method... that's provides what you > are looking for. Excellent. > >> The goal is to just have the qsar.xml (see blog post in top) and be >> able to download all molecules from a public repository, > > InChIs... It's easy to make Bioclipse aware of where to download/get > what for a certain InChI... Yes, that would be another option in addition to REST ID's. How would one go about and do that? What's the status of Inchis in Bioclipse? I have created a wiki page about this, see: http://wiki.bioclipse.net/ index.php?title=QSAR_Project_with_automated_builds Cheers, .../Ola > >> and automatically calculate the descriptors from within Bioclipse. > > Egon > > -- > ---- > http://chem-bla-ics.blogspot.com/ |
From: Egon W. <ego...@gm...> - 2008-01-07 15:06:11
|
On Jan 7, 2008 4:00 PM, Ola Spjuth <ola...@fa...> wrote: > On Jan 2, 2008, at 22:49 , Egon Willighagen wrote: > > On Dec 21, 2007 12:38 PM, Ola Spjuth <ola...@fa...> wrote: > >> 1) > >> Is it possible to publish molecules to a public repository and get > >> REST ID's to them (something like below)? > >> http://www.mol.repo.org/molecules/cml/id?abc234 > >> http://www.mol.repo.org/molecules/mol2/id?abc234 > > > > PubChem? > > I don't think it's currently possible in Pubchem. I strongly suggest the RDF IDs I use for rdf.molecules.net, which uses the InChI, e.g.: http://rdf.openmolecules.net/?InCh=1/CH4/h1H4 > Rajarshi: If you could add REST IDs to your Pubchem via CID that > > InChIs... It's easy to make Bioclipse aware of where to download/get > > what for a certain InChI... > > Yes, that would be another option in addition to REST ID's. > How would one go about and do that? What's the status of Inchis in > Bioclipse? For 32bit Windows and Linux the JNI wrapper by Sam can be used. In other cases, the Java version Rich created of OpenBabel could be an option. > I have created a wiki page about this, see: http://wiki.bioclipse.net/ > index.php?title=QSAR_Project_with_automated_builds Check out http://sourceforge.net/projects/jexcelapi Egon -- ---- http://chem-bla-ics.blogspot.com/ |
From: Rajarshi G. <rg...@in...> - 2008-01-07 15:12:07
|
On Jan 7, 2008, at 10:00 AM, Ola Spjuth wrote: > Egon: Yes. > > I will need REST URL's that define the descriptors, What do you mean by this? I have provided a web service to get descriptor specifications and REST URL to the desc specs for all descriptors is http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications Does this work for you? > If we share the same REST ID's then Rajarshi's Web services could > be plugged into Bioclipse and be available as a separate descriptor > provider (in addition to CDK and JOELib). Should be easy to set up. I'm a little unclear on what you mean by REST ID's. My understanding is that a REST URL defines a resource which may be a value (descriptor value in the case) or links to other resources. So by doing http://rguha.ath.cx/~rguha/cicc/desc/descriptors/CCCC you get a set of links to other resources for the specified molecule. What is the REST ID here? > Link: http://wiki.bioclipse.net/index.php? > title=QSAR_Project_with_automated_builds THis seems to be an empty page ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- Paper or plastic? Not 'Not paper AND not plastic!!' -- Augustus DeMorgan in a grocery store |
From: Ola S. <ola...@fa...> - 2008-01-07 15:50:36
|
Works for me... <http://wiki.bioclipse.net/index.php? title=QSAR_Project_with_automated_builds> On Jan 7, 2008, at 16:12 , Rajarshi Guha wrote: > >> Link: http://wiki.bioclipse.net/index.php? >> title=QSAR_Project_with_automated_builds > > THis seems to be an empty page > |
From: Rajarshi G. <rg...@in...> - 2008-01-07 15:32:36
|
On Jan 7, 2008, at 10:00 AM, Ola Spjuth wrote: > > On Jan 2, 2008, at 22:49 , Egon Willighagen wrote: > >> On Dec 21, 2007 12:38 PM, Ola Spjuth <ola...@fa...> >> wrote: >>> 1) >>> Is it possible to publish molecules to a public repository and get >>> REST ID's to them (something like below)? >>> http://www.mol.repo.org/molecules/cml/id?abc234 >>> http://www.mol.repo.org/molecules/mol2/id?abc234 >> >> PubChem? > > I don't think it's currently possible in Pubchem. > > Rajarshi: If you could add REST IDs to your Pubchem via CID that > would be great! Would it be possible to go via InChi as well? And > what would it return, an SDF or a SMILES string or what? Certainly doable, but what do you want returned for a given InChI? Since we provide multiple databases here I've gone ahead and created a REST interface for our 3D structure DB. For example the URL http://rguha.ath.cx/~rguha/cicc/db/pub3d/InChI=1/C9H8O4/c1-6(10) 13-8-5-3-2-4-7(8)9(11)12/h2-5H,1H3,(H,11,12)/f/h11H returns the 3D structure of aspirin in SD format. So I can easily see the query from the 3D DB, but since PubChem holds more info, what would you want from it? ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- Entropy requires no maintenance. -- Markoff Chaney |
From: Rajarshi G. <rg...@in...> - 2008-01-07 15:44:29
|
On Jan 7, 2008, at 10:32 AM, Rajarshi Guha wrote: > > Since we provide multiple databases here I've gone ahead and created > a REST interface for our 3D structure DB. For example the URL > > http://rguha.ath.cx/~rguha/cicc/db/pub3d/InChI=1/C9H8O4/c1-6(10) > 13-8-5-3-2-4-7(8)9(11)12/h2-5H,1H3,(H,11,12)/f/h11H > > returns the 3D structure of aspirin in SD format. CID also works, so that http://rguha.ath.cx/~rguha/cicc/db/pub3d/2244 gives the 3D structure for aspirin (CID = 2244) ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- "I'd love to go out with you, but my favorite commercial is on TV." |
From: Ola S. <ola...@fa...> - 2008-01-07 15:49:43
|
On Jan 7, 2008, at 16:32 , Rajarshi Guha wrote: > On Jan 7, 2008, at 10:00 AM, Ola Spjuth wrote: >> On Jan 2, 2008, at 22:49 , Egon Willighagen wrote: >>> On Dec 21, 2007 12:38 PM, Ola Spjuth <ola...@fa...> >>> wrote: >>>> 1) >>>> Is it possible to publish molecules to a public repository and get >>>> REST ID's to them (something like below)? >>>> http://www.mol.repo.org/molecules/cml/id?abc234 >>>> http://www.mol.repo.org/molecules/mol2/id?abc234 >>> >>> PubChem? >> >> I don't think it's currently possible in Pubchem. >> >> Rajarshi: If you could add REST IDs to your Pubchem via CID that >> would be great! Would it be possible to go via InChi as well? And >> what would it return, an SDF or a SMILES string or what? > > Certainly doable, but what do you want returned for a given InChI? > > Since we provide multiple databases here I've gone ahead and > created a REST interface for our 3D structure DB. For example the URL > > http://rguha.ath.cx/~rguha/cicc/db/pub3d/InChI=1/C9H8O4/c1-6(10) > 13-8-5-3-2-4-7(8)9(11)12/h2-5H,1H3,(H,11,12)/f/h11H > > returns the 3D structure of aspirin in SD format. That's very nice! I will use that as an example. > > So I can easily see the query from the 3D DB, but since PubChem > holds more info, what would you want from it? I don't know, we'll have to work it out as we go. Basically everything in the most general way that is needed to calculate descriptors, but this is really not my area of expertise. Best would IMO be to get CML with 2D and 3D coordinates and annotations. > > > > ------------------------------------------------------------------- > Rajarshi Guha <rg...@in...> > GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE > ------------------------------------------------------------------- > Entropy requires no maintenance. > -- Markoff Chaney > > |
From: Rajarshi G. <rg...@in...> - 2008-01-07 16:01:34
|
On Jan 7, 2008, at 10:51 AM, Ola Spjuth wrote: >> Since we provide multiple databases here I've gone ahead and >> created a REST interface for our 3D structure DB. For example the URL >> >> http://rguha.ath.cx/~rguha/cicc/db/pub3d/InChI=1/C9H8O4/c1-6(10) >> 13-8-5-3-2-4-7(8)9(11)12/h2-5H,1H3,(H,11,12)/f/h11H >> >> returns the 3D structure of aspirin in SD format. > > That's very nice! I will use that as an example. > >> >> So I can easily see the query from the 3D DB, but since PubChem >> holds more info, what would you want from it? > > I don't know, we'll have to work it out as we go. Basically > everything in the most general way that is needed to calculate > descriptors, but this is really not my area of expertise. Best > would IMO be to get CML with 2D and 3D coordinates and annotations. Hmm, our Pub3D DB provides coverage for 99% of Pubchem, but currently only contains a single conformer (not necessarily lowest in energy) of a structure. Also we don't store the 2D coords of PubChem structures - but that's easily obtainable using the CDK. Regarding annotations - that brings us back to what we want from the DB. We store a number of fields (various SMILES, charge, heavy atom count,synonyms etc) - would you want all of it one CML file (basically a whole Pubhchem record)? Or as individual values? ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- Entropy isn't what it used to be. |
From: Ola S. <ola...@fa...> - 2008-01-07 15:53:07
|
I think we are talking about the same thing but different terminology (expecially me). I am new to REST but I'll try to explain what my intentions are: In REST, each resource has a URL which uniquely identifies the resource. Here, I'd like to have a URL that identifies a molecule/ descriptor, and by calling it with HTTP GET it will return the actual molecule/descriptor. So when I say REST ID I should probably say Resource URL. Maybe even, in the xml, use URL instead of ID? On Jan 7, 2008, at 16:12 , Rajarshi Guha wrote: > > What do you mean by this? I have provided a web service to get > descriptor specifications and REST URL to the desc specs for all > descriptors is > > http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications > > Does this work for you? Yes, this is good. Do individual descriptorSpecifications also have URL's? Like: http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications/ org.openscience.cdk.qsar.descriptors.molecular.ALOGPDescriptor will return: <DescriptorSpecification class="org.openscience.cdk.qsar.descriptors.molecular.ALOGPDescriptor"> <SpecificationReference> http://www.blueobelisk.org/ontologies/chemoinformatics-algorithms/ #ALOGP </SpecificationReference> <ImplementationTitle> org.openscience.cdk.qsar.descriptors.molecular.ALOGPDescriptor </ImplementationTitle> <ImplementationIdentifier>$Id: ALOGP-MR descriptor $</ ImplementationIdentifier> <ImplementationVendor>The Chemistry Development Kit</ ImplementationVendor> </DescriptorSpecification> And: http://www.blueobelisk.org/ontologies/chemoinformatics-algorithms/ #ALOGP would be the URL of the descriptor. Egon: This URL does not work. It should return the XML description for this descriptor, right? |
From: Rajarshi G. <rg...@in...> - 2008-01-07 16:12:47
|
On Jan 7, 2008, at 10:55 AM, Ola Spjuth wrote: > I think we are talking about the same thing but different > terminology (expecially me). > I am new to REST but I'll try to explain what my intentions are: > > In REST, each resource has a URL which uniquely identifies the > resource. Here, I'd like to have a URL that identifies a molecule/ > descriptor, and by calling it with HTTP GET it will return the > actual molecule/descriptor. So when I say REST ID I should probably > say Resource URL. Maybe even, in the xml, use URL instead of ID? Aah OK. I'm fine with either terminology, but since it's all URL's it sounds nice to talk about REST URL's. However, on the wiki page that you linked to you have <descriptor id="http://www.blueobelisk.org/ontologies/ chemoinformatics-algorithms/#xlogP" name="XlogP"> <parameter key="chechAromaticity" value="true"> <parameter key="salicylFlag" value="false"> <descriptorImplementation id="http://rguha.ath.cx/~rguha/cicc/ desc/descriptors/ org.openscience.cdk.qsar.descriptors.molecular.XLogPDescriptor" /> </descriptor> The URL used in the id attribute of the descriptorImplementation is not a valid URL (you'll get a HTTP 405 error) since it does not have a SMILES at the end. I'm think of converting the descriptors resource such that visiting the URL you have, would return the details of the specification rather than a 405 error. That way we wouldn't need a separate specification resource at all. > > On Jan 7, 2008, at 16:12 , Rajarshi Guha wrote: >> >> What do you mean by this? I have provided a web service to get >> descriptor specifications and REST URL to the desc specs for all >> descriptors is >> >> http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications >> >> Does this work for you? > > Yes, this is good. Do individual descriptorSpecifications also have > URL's? Like: > > http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications/ > org.openscience.cdk.qsar.descriptors.molecular.ALOGPDescriptor Yes, this now works. Also I have changed the result of the URL http://rguha.ath.cx/~rguha/cicc/desc/descriptors/specifications which rather than returning the specifications directly, returns links which will return individual specifications. THis is in line with the REST interface for descriptors ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- I saw Elvis. He sat between me and Bigfoot on the UFO. |