PDB to SDF

How to convert a ligand PDB to SDF?

To convert PDB to SDF, isolate the ligand in its own PDB file, read it with RDKit, Open Babel, or PyMOL, and write it out as SDF. The program perceives bonds and bond orders from the coordinates and CONECT records, because PDB does not store them reliably. You can do it online with this converter, with the obabel command, or by saving a selection in PyMOL.

Online with the PDB to SDF converter

RDKit reads the coordinates and connectivity, writes an SDF, and reports the perceived SMILES and atom count, including for peptide ligands.

1. Extract the intended ligand or peptide into its own .pdb file and upload it.
2. Leave Infer bonds from proximity and Remove hydrogens on, or turn either off.
3. Run the converter and compare the reported SMILES with the ligand you expect.
4. Download the SDF from Files.

On the command line with Open Babel

obabel ligand.pdb -O ligand.sdf

Open Babel takes both formats from the file extensions. Add -h to add hydrogens to a ligand taken from a crystal structure. Open Babel perceives bond orders with its own rules, so when its SDF and RDKit's disagree on a ligand, the PDB is ambiguous and the chemistry needs checking against a reference.

In PyMOL

PyMOL can save a ligand straight out of a complex. Load the structure, then save only the ligand's residue name:

load complex.pdb
save ligand.sdf, resn ATP

Replace ATP with the ligand's three-letter code from the HETATM lines.

Sample input and output

Input .pdb:

COMPND    ethanol
HETATM    1  C1  UNL     1      -0.953   0.048   0.042  1.00  0.00           C  
HETATM    2  C2  UNL     1       0.488  -0.322  -0.189  1.00  0.00           C  
HETATM    3  O1  UNL     1       1.277   0.325   0.738  1.00  0.00           O  
HETATM    4  H1  UNL     1      -1.357   0.734  -0.733  1.00  0.00           H  
HETATM    5  H2  UNL     1      -1.593  -0.863   0.022  1.00  0.00           H  
HETATM    6  H3  UNL     1      -1.071   0.596   1.011  1.00  0.00           H  
HETATM    7  H4  UNL     1       0.787  -0.141  -1.236  1.00  0.00           H  
HETATM    8  H5  UNL     1       0.590  -1.426  -0.044  1.00  0.00           H  
HETATM    9  H6  UNL     1       1.832   1.048   0.389  1.00  0.00           H  
CONECT    1    2    4    5    6
CONECT    2    3    7    8
CONECT    3    9
END

Actual .sdf output:

ethanol
     RDKit          3D

  3  2  0  0  0  0  0  0  0  0999 V2000
   -0.9530    0.0480    0.0420 C   0  0  0  0  0  0  0  0  0  0  0  0
    0.4880   -0.3220   -0.1890 C   0  0  0  0  0  0  0  0  0  0  0  0
    1.2770    0.3250    0.7380 O   0  0  0  0  0  0  0  0  0  0  0  0
  1  2  1  0
  2  3  1  0
M  END
$$$$

The blocks show the complete input and output for one ethanol molecule. Coordinates and chemistry shown here come from this conversion run.

Input

Settings

Results

Downloads are available from Files. Inspect both the structure preview and the chemical content of the downloaded file.

PDB coordinates do not fully define chemistry

PDB is primarily a coordinate format. CONECT records can provide connectivity; proximity bonding lets RDKit infer additional bonds from atom distances. Neither guarantees the intended bond orders, aromaticity or charges for an arbitrary ligand.

Prefer an original SDF or chemically complete SMILES when available. Successful conversion means a molecular file was written, not that perceived chemistry matches the intended ligand. Every result warns about this limitation. Compare the reported SMILES with a chemically complete reference.

By default, explicit hydrogens are removed during PDB reading. Turn Remove hydrogens off to retain input hydrogens. This does not add missing hydrogens, predict protonation or repair coordinates.

Extract a ligand before conversion

Supply the intended ligand or peptide as its own PDB. Amino-acid residue names do not prevent conversion. The converter reads the supplied structure without extracting a ligand; waters or unrelated molecules can otherwise enter the result.

For docking preparation from known chemistry, use SMILES to PDBQT. With a complete 3D SDF, SDF to PDBQT avoids interpreting chemistry from PDB.

FAQ

How do I convert PDB to SDF with Open Babel?

Run obabel ligand.pdb -O ligand.sdf, adding -h if the ligand needs explicit hydrogens. Extract the ligand first, because Open Babel converts every atom in the file, including protein and water.

Can I convert PDB to SDF in PyMOL?

Yes. Load the PDB and run save ligand.sdf, resn ATP, replacing ATP with the ligand's residue name, or use File, then Export Molecule, and choose SDF. Saving a selection extracts the ligand and converts it in one step.

Can I convert a complete protein PDB to SDF?

RDKit decides whether the supplied PDB can be parsed as a molecule. This converter does not extract ligands, repair a protein or prepare it for docking. For a ligand result, upload the isolated ligand or peptide rather than a protein complex.

Does PDB to SDF recover missing bond orders?

It writes RDKit's perceived chemistry. Missing bond-order or charge information cannot be guaranteed without a complete chemical reference.

Why are there fewer atoms in the SDF?

With Remove hydrogens on, explicit hydrogens are removed during reading. Heavy atoms remain in the parsed structure.