Sorry for dragging my feet on this one...
I've implimented the following fix:
relax_data.read() no-longer attempts to deselect residues.
This means that all residues in the sequence will be selected,
irrespective of whether or not they contain data, unless the user says
so, or minimise(), grid_search(), or calc() are called.
Over-fitting is prevented on calls to these functions by de-selecting
residues with k > n. This overfit_deselect() function treats mf, jw,
and relax_fit run types appropriately, and also checks for structural
data where required.
The only draw-back I see with this is that the user probably won't
expect minimise() to change residue selections. Not too much of a big
deal though...
The other fix suggested was that relax_data.read() re-select residues
when data is found. I don't like this option because the user might do
unselect.res() before doing relax_data.read(). Then having
relax_data.read() do re-selections could override the users selctions
ie. relax_data.read() has no way of knowing if a residue is de-selected
because a of previous relax_data.read() call or because of an explicit
unselect.res()
The fix as I've implemented it passes the test-suite as well as a
barrage of relavant sample scripts (though see bug #5698)
Chris
Re: Resolution to bug #5501, "residue deselection problem on relax_data.read()".