I did it before by editing the pdb file. You can do the editing by hand or write a script to do it.

You can do that with Modeller (http://salilab.org/modeller/)

You should have a clear idea about which is the secondary structure that the missing residues adopt before trying to model them. This can be done by modeling these residues by similarity modeling with a template that has a high degree of homology with your target molecule. For this sake you could use freely available software like SPDB viewer.

you can do with manually as well as softwares. In manually, you open the pdb file in wordpad and manually enter the residues and connection . and you also using the spdbv, modeller, Pdbeditor ,etc.which easy for you to editing the pdb script.

If you are talking about missing loops, then yes, you can simply use ModLoop. First you can add your missing residues manually in your PDB file (e.g. if it is a histidine, you can just copy any histidine residue atoms and use the dummy coordinate) followed by refinement using ModLoop server to fix the coordinate.

You can use PDBFIXER with the original information from your *.PDB file.

I suggest you use Modeller, as recommended by Miguel Mompeán

I haven't used other software for this purpose because I'm pretty happy with Modeller.

You may explore this tool: pdbfixer

https://github.com/openmm/pdbfixer

Modeller is not applicable for dna