SV: Re: [Cdk-user] Where is the DBReader class

[<cla...@ya...>]

Christoph,
 I have added some comments below:
 
 before we go on, we should clarify what you are planning to do with Chemical Inventory (CI). Some design decisions that we made for NMRShiftDB might not neccessarily be suited for CI. 
 
 The main idea is to create a "full med chem support package". Currently CI is mainly used as a reagent database, but the containers could also hold in house synthesised compounds. These compounds need to be associated with (at least) in vitro pharmacology results as well as calculated phys/chem parameters in order to give the Med Chem's a good SAR table basis for further design.
 
 So we are mainly in need of structure searching but also parameter calculations based on the structure.
 
 The splitting of molecules into atom and bond tables on the SQL level is probably not what you need and want and it might make things slow.
 
 No, it doesn't look like it. But that will of cause make the coding easier.
 
 If the main purpose is searching for structures, it will be enough to store the molecule as, say, a CML snippet in on table, together with fingerprint and some additional information (name, gross formula, etc.). Most structure databases work like this. 
 
 Depending on your answer with regards to the parameter calculations I think we will just need a single molecule table with CML. smiles, fingerprints and the trivial stuff.
 What format is needed if we want to be able to export sd-files from the database using CDK?
 
 However, I have asked my coworker Stefan to dig into the NMRShiftDB code and show us the relevant snippets for structure searching.
 
 Very nice!
 
 BTW, did you already discover the "CDK News", our quartly newsletter. It will have a lot of interesting articles that answer questions that you might have.Take a look at http://almost.cubic.uni-koeln.de/cdk/cdk_top/cdk_news/ and get the various pdf's.
 
 Yes, but i can't say that I have read it all. But will take another look and read. thanks.
 
 Another valuable source of information is the org.openscience.cdk.test package of CDK, where each class in CDK, such as the fingerprinter, has its test class. These tests show, how to use and setup each of the functionalities provided by the CDK. Take a look, for example, at the test.fingerprint.FingerprinterTest.java file to learn how to read a molfile and make a fingerprint from it.Equally, test.smiles.SmilesGeneratorTest shows you more on this side.
 
 Sounds like a very good idea. I didn't know about this part.
 
 cheers
 
 claus