From: Andrew Dalke (dalke_at_dalkescientific.com)
Date: Wed Nov 12 2003 - 13:29:15 CST

Vlad Cojocaru:
> I noticed that the selection "nucleic" does not recognize the terminal
> residues from a NA molecule. The 5' end residues that have no
> phosphate (RN5 and DN5 named in amber...N is any nucleotide) are not
> recognized? Could this be fixed and if yes could you give me a hint
> how?

That's my doing, assuming no one has changed the code
in the last 6 years (has it been that long?!).

The 'protein' and 'nucleic' commands use a simple, almost
brain-dead way to figure out if a residue is of the given
type. The 'protein' one looks for a residue which has
atoms 'C', 'N', 'O', and 'CA'. (It also allows OT1 and
OT2 as terminal oxygens, as I recall.). I don't even
think it checks to see if those are backbone atoms, but
my memory is fuzzy on that. I know it doesn't check if
the atoms are in the correct order.

The 'nucleic' one is not so simple. VMD again checks
for the atom names both for the sugars and the phosphate.
(In this case it looks for both C3' and C3* because
older PDB files used * instead of '.) However, the
terminal bases don't have a phosphate so there are
enough atoms to meet VMD's criteria to be 'nucleic'.

I recall I tried reducing the number of atoms needed
for a match, but that caused false positives with
other molecules. As it is, ATP is included as
'nucleic', but some people thought that was a feature
so I didn't try to fix it further.

It could be fixed by having a more complicated
detection scheme. I know what I wrote was very simple
minded. I figured if it was bad then people would
complain and I would fix it, but few did and I didn't.

                                        Andrew
                                        dalke_at_dalkescientific.com