PDB Sources

Some ways of obtaining a coordinate file for the molecule(s) of interest are:

• Protein Data Bank
• Hetero-Compound Information Centre: Uppsala - information about hetero-compounds encountered in files from the Protein Data Bank (PDB), with each entry providing coordinate files (PDB and text files), visualisation files (Chime and VRML), dictionary files (X-PLOR/CNS, O, TNT), links to off-site databases and servers, and miscellaneous files and information.
• Het-PDB Navi - small molecules found in the PDB
• Construct yourself using molecule building software, e.g. SYBYL, PyMOL or Accelrys Discovery Studio
• PRODRG - sketch your small molecule in a simple text editor and coordinate / topology files are returned.
• ArgusLab could build your protein and/ or import it from a known PDB file database. If you choose to build your protein, it will export it in a PDB format. If you download one from a database, obviously you can save it in a PDB format. It is also specifically designed to make building proteins simpler (not simple - just simpler) with a built-in library of amino acids in fragment forms that can be connected to create the protein of your choice along with side chain fragments, solvation species, etc.
• OpenBabel intercoverts most coordinate file types so, you could build your molecule with any of the 3D model builders that serve as front ends to Quantum Chemistry programs. Export it as an XYZ file and then use Openbabel to convert it to PDB format. Some will even export PDB formatted files.
• Avogadro open-source cross-platform molecule builder, includes support for peptides
• xLeap - part of the AmberTools distribution.  Can build small molecules and peptides.
• MemGen - generates multi-component lipid membranes for use in atomistic molecular dynamics simulations with any forcefield, lipid and MD software

If you are missing some heavy atoms, then some suggestions from gmx-users were: Swiss-PDBviewer, WhatIf, Loopy, SCWRL3 (or SCWRL4), coot, maybe PyMOL, or a text editor. Consider adapting coordinates from a similar residue, or building and replacing the residue.

If you are missing some hydrogens, then GROMACS can help. Either pdb2gmx will build them in if suitable .hdb and .rtp entries are present, or perhaps use protonate.