Difference between the Radial Distribution Function and number density


I want to calculate the number density of an atom of my ligand as function of distance from protein heavy atoms. In the manual, it says the following:

The RDFs are normalized by 1) average number of positions in -ref (the number of groups with -surf ), 2) volume of the bin, and 3) average particle density of -sel positions for that selection. To change the normalization, use -norm :

rdf : Use all factors for normalization. This produces a normal RDF.
number_density : Use the first two factors. This produces a number density as a function of distance.

I wanted to know what the difference between the number density and the normal RDF in Gromacs. Apologies if this may sound like a simple question. Any guidance will be much appreciated.



Hi Akash,

Thanks for posting!

In your case it should be the same, because you only look at one atom, if I understand you correctly. If you had two ligands, then the number density would be twice as much as the “normal” rdf and so on…


Thank you for your response. I have hundreds of copies of my ligand in my system, so my understanding on what you said , the number density in this case will be “normal rdf” multiplied by 100 (100 time as much)? Is this if I only choose 1 atom?

Alternatively, what would be the case if I took the center of mass of all my ligands (using -seltype mol_com)? If I wanted to calculate the number density would I then normalise (e.g. -norm number_density)?

Hi Akash,

you will get a factor for every -sel positions for that selection. That is, if you look at the center of mass of hundred molecules, you will get one-hundred -sel positions, similarly looking at one atom each of hundred molecules will give you the same factor.