[phenixbb] rotamers

Ed Pozharski epozh001 at umaryland.edu
Thu Mar 24 08:23:17 PDT 2011

> >   I have seen
> > some quite convincing evidence since that the disordered side chains do
> > have a detectable effect on the rest of the model, and thus leaving them
> > in makes a model worse.
> I've seen both.

What do you mean?  That removing disordered side chains affects the rest
of the model too?  But surely removing atoms not found in the density
will reduce overall model error.

> > Other half of the argument was of semantic
> > nature and referred more to replacing disordered residues with alanines,
> > which is silly
> It is silly but honest. 

Again, not sure what you mean.  Tyrosine is a tyrosine even if positions
of some atoms are unknown/omitted.  If I rename it to alanine, then I am
claiming it's an alanine, which is wrong.  (Of course, one can be wrong
and honest at the same time).

> I agree, making occupancy zero is kind of abusing it in order to say "I 
> do not see this atom". 

No, that is not what occ=0.00 says.  100% occupancy means in the
standard model that an atom has an average position at (x,y,z) and moves
around (dynamically and statically) in a harmonic fashion with the
amplitude defined by B-factor.  50% occupancy means that it is found 50%
of time in the "gaussian vicinity" of (x,y,z), and 50% of time it's
somewhere else (in the second location present as alternate conformation
or, if the latter is absent, everywhere else as allowed by geometry).
10% means it is only found at (x,y,z) 10% of the time, etc.  What
happens at 0% depends on one's take on continuity of the occupancy as a
descriptor of physical reality.  Unless we redefine occupancy of
precisely 0.00 as a special case that actually means "I have no idea
where this atom is other than what I can deduce from geometry and by the
way, the coordinates provided are completely meaningless" then by
continuity it means that the probability to find this atom at (x,y,z) is
precisely 0.  Which, for a disordered side chain, is most likely

> Finally, when we model 4-6A or so resolution data why we stick atoms 
> into those tubes of density? Do these densities really tell you where 
> that specific atom or often even residue is?

With low precision, but yes.  A correct analogy here is placing extra
residues when there is no tube of density, e.g. missing loops and/or
termini.  But nobody does that.  We just stick with partial model.  Why
should side chains be any different?


"I'd jump in myself, if I weren't so good at whistling."
                               Julian, King of Lemurs

More information about the phenixbb mailing list