From: Rajarshi G. <raj...@gm...> - 2011-11-28 03:07:19
|
On Mon, Nov 14, 2011 at 7:31 AM, Jochen Schreiber <jo...@go...> wrote: > The databases that i could download from the pubchem ftp for the substances > havnt any 3D informations. Pubchem has a 3D version, though they only store a 1 or 10 conformers. See ftp://ftp.ncbi.nlm.nih.gov/pubchem/Compound_3D > So the following code (which i test for iterating > through all informations) will fail: Note that if you want to read through an SD file that contains multiple conformers (numbered contiguously) then you should be using the IteratingMDLCoformerReader and ConformerContainer classes > 1. Converting structure files into sdf (with babel) Do you mean generate 3D structures? > 2. CDK for creating the query.xml for CDKPSearch You could also construct the query by hand (probably easier) > 3. CDKPSearch for searching structures in the database (pubchem) > 4. CDK for analyse the results (output the results etc.) > Is this the right way? This is certainly one way, though you might want to consider whether 10 conformers per structure is appropriate for a pharmacophore search -- Rajarshi Guha NIH Chemical Genomics Center |