Hi Paul, That's fixed in the GitHub master trunk of the RDKit, while it doesn't work with the released version (I know because I fixed that bug myself!). Most SDF files in the RCSB feature right-aligned elements, which is indeed allowed by the CT file specs.
Cheers, p. > On 10 Sep 2016, at 21:17, Paul Emsley <[email protected]> wrote: > > > Thank you Paolo, that's very useful. > > But does that file work for you? > > >>> import rdkit > >>> from rdkit import Chem > >>> ms = [x for x in Chem.SDMolSupplier('example.sdf')] > [21:04:24] > > **** > Post-condition Violation > Element '' not found > Violation occurred on line 91 in file > /Users/pemsley/autobuild/b-e-l/rdkit-Release_2015_03_1/Code/GraphMol/PeriodicTable.h > Failed Expression: anum>-1 > **** > > [21:04:24] Unexpected error hit on line 5 > > Comparing the heads: > > % head Ligands_noHydrogens.sdf > 1CX2_S58_A_701 > RCSB PDB01141507103D > Coordinates from PDB:1CX2:A:701 Model:1 without hydrogens > 26 28 0 0 0 0 999 V2000 > 27.5570 21.7340 16.8580 C 0 0 0 0 0 0 0 0 0 0 0 0 > 26.3220 22.0290 16.3660 C 0 0 0 0 0 0 0 0 0 0 0 0 > 27.3340 21.1490 18.2120 C 0 0 0 0 0 0 0 0 0 0 0 0 > 28.3530 20.7370 19.2320 C 0 0 0 0 0 0 0 0 0 0 0 0 > 24.0280 21.4680 17.1130 C 0 0 0 0 0 0 0 0 0 0 0 0 > 23.4160 20.4550 17.8620 C 0 0 0 0 0 0 0 0 0 0 0 0 > > % head S58_ideal.sdf > S58 > -ISIS- 3D > > 37 39 0 0 0 0 0 0 0 0 0 > 1.3830 0.0050 -2.9870 C 0 0 0 0 0 > 0.8620 0.0130 -1.7240 C 0 0 0 0 0 > 0.3190 -0.0060 -3.8900 C 0 0 0 0 0 > 0.4400 -0.0140 -5.3920 C 0 0 0 0 0 > -1.4220 -0.0020 -0.7940 C 0 0 0 0 0 > -2.5780 0.7620 -0.8830 C 0 0 0 0 0 > > The element column has shifted rightwards 2 spaces in > Ligands_noHydrogens.sdf. Weird? From this output and presuming that if RDKit > can't read it then it's wrong, it seems to me that RCSB are distributing > non-conformant sdf files. > > Do you agree? > > Paul. > > >> On 10/09/2016 18:58, Paolo Tosco wrote: >> Hi Paul, >> >> SDF files in the RCSB database do store PDB atom names, have a look at >> >> http://www.rcsb.org/pdb/download/downloadLigandFiles.do?ligandIdList=S58&structIdList=1CX2&instanceType=all&excludeUnobserved=false&includeHydrogens=false >> >> >> for an example. >> >> Atom names stored in this fashion are read in by the RDKit and can be >> accessed as the >> "molFileAlias" atom property; they can also be stored accordingly. See >> >> https://gist.github.com/ptosco/6e4468350f0fff183e4507ef24f092a1#file-pdb_atom_names-ipynb >> >> for an example. >> >> Cheers, >> p. >> >>> On 10/09/2016 18:33, Paul Emsley wrote: >>> Hi RDKiters, >>> >>> Is it possible to store (PDB) atom names in sdf files? If so, how is this >>> done? (It was not >>> clear to me after reading CTFile.pdf.) >>> >>> (I hope that this question is allowed.) >>> >>> Thanks, >>> >>> Paul. >>> >>> ------------------------------------------------------------------------------ >>> _______________________________________________ >>> Rdkit-discuss mailing list >>> [email protected] >>> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss >> > ------------------------------------------------------------------------------ _______________________________________________ Rdkit-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
