[gmx-users] How to choose appropriate size for a sliding average for smoothing distances between molecule types

[Anna Duncan]

 I have performed a simulation of a protein in a lipid bilayer and I
 want to look at the lengths of time that a particular lipid type, lipid
 type A, say, remains within a certain distance of the protein.

 I have calculated the distances between lipid type A and the protein.
 Obviously, the trajectories of each particle are not smooth, so neither
 are the distances between atoms of lipid type A and the protein.
 Because I want to look at a continuous amounts of time that lipid type 
 A
 molecules are proximal to the protein, it seems to make sense to smooth
 the distances, using xmgrace, for instance.  Does this make sense to
 you?  If so, how should I go about determining by how much to smooth 
 the
 distances, or indeed the atom trajectories?  Is it worth looking at the
 velocity autocorrelations or is this a slightly different question?


 Many thanks,

 Anna