On 11/04/2013 15:45, Pascal Muller wrote: > Hi, > > In a similarity search, obabel consider another molecule than the > reference I'm giving. > > My library being test.smi: > > C1(=CC2=C(C(C(C2)(C#N)C#N)(CC(=O)C)C#N)C(=O)N)C(=O)NC(=O)S1 A > C1(=Cc2c(oc3c(cc(cc3)Br)c2=O)C(=O)N)C(=O)NC(=O)S1 B > C1(=CC2=C(C(C(C2)(C#N)C#N)(C2CCCC2=O)C#N)C(=O)N)C(=O)NC(=O)S1 C > > after conversion in fs format: > obabel test.smi -ofs > > The command: > obabel test.fs -SA.smi -ofpt -at0.0 > is giving: > >> C >> A Tanimoto from C = 1 > Possible superstructure of C >> B Tanimoto from C = 0.278689 > 3 molecules converted > > A is _not_ the reference (first line). obabel is considering C instead. > > > B as reference is ok: > obabel test.fs -SB.smi -ofpt -at0.0 >> B >> C Tanimoto from B = 0.278689 >> A Tanimoto from B = 0.278689 > 3 molecules converted > > C as reference is OK, and is producing the same result as if taking A > as reference. > > > Changing the order in test.smi (e.g. C, A, B), using C as reference > returns A in the first line. Using A as reference is OK. > > I have 43 similar cases out of my 1579 compounds library. > Any ideas apart from shuffling the smiles file? :)
fs and fpt formats are not really intended to be used together. If you want to find the 10 most similar molecules to molecule A in a *large* dataset use obabel test.fs -osmi -s A.smi -at10 OR to show similarities from A to each molecule in a *small* dataset obabel A.smi test.smi -ofpt (The target molecule is the first molecule that is output.) You shouldn't need to use -at0.0 (even when its bugs are corrected, see next post). Chris ------------------------------------------------------------------------------ Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter _______________________________________________ OpenBabel-discuss mailing list OpenBabel-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openbabel-discuss
