Hi Bryan:

I think the question is hard to answer because the idea of "bad" phases is
not particularly well-understood (at least by me).

Good phases give you a nice map.

"Weak phases" give you a map that is weakly interpretable, but often can
be improved by solvent flattening, NCS averaging, and so forth.

"Bad phases" I would think have something inherently wrong with them, such
as some systematic bias that would prevent you from getting to the right
answer.

I once gleefully showed my postdoctoral advisor phasing statistics for a
macromolecule I had just solved.  The overall phasing power was 0.6 for
one derivative and 0.5 for another.  Phasing power is roughly\

Sum |F(PH) - F(P)|/deltaE

where deltaE is lack-of-closure error, so it is typically looked at as an
average signal to noise value.

I thought if the lack of closure error on average was roughly twice my
heavy atom signal, the phases should be too weak to give a decent map. It
turns out I can make a really nice map with only one of derivatives
(either one).

My postdoctoral advisor retorted "When I introduced that statistic" (I
knew I was doomed at that point), "I never meant to suggest that a phasing
power less than 1.0 meant the phases were no good."

I think the reason it works is because the lack of closure error is nearly
random and the signal is not, and that, together with a low initial figure
of merit and a high solvent content, it meant that the weak phases were
good enough to make a nice map.

I've also heard that pretty much anything better than pure random noise
when combined with lots of NCS (viruses) can yield nearly perfect phases.

The main thing is to avoid bad (meaning nefariously biased) phases.

Bad partial models, bad molecular replacement solutions, bad heavy atom
constellations, and anything else involving atoms in wrong positions is
much more problematic than a correct heavy atom model with a weak phasing
power.

Low initial FOM in many senses are good, because density modification has
more latitude to help.

Bill




Bryan W. Lepore wrote:
> general question - perhaps the fundamental question -
>
> for anyone who had "weak/poor/bad" phases from some source, that were
> later actually used to solve a structure when combined w/ another source -
> HOW bad were the worst phases on their own, in terms of resolution, FOM,
> CC, e-density, (any other numbers)?  what was MOST important in knowing
> the phases would help (presumably e-dens.).
>
> i.e, was it only when relatively "better" phases gave any interpretable
> density that it was known that the "bad" phases would help?
>
> -bryan
>