> I can't find the paper I read that suggested that in practice the
> reverse correlation can be true.
Of course it can. Those are conditional probability distributions
that tell you what is most likely the case. The distributions
tell you just how possible that is with its corresponding probability.
No contradiction here at all. One interesting case Dale Tronrud
reported to me where he got a zero probability was in a previous
edition of the calculator where I rounded the printout too
aggressively. Then it turned out to be just as expected :-)
On top of it, there goes nothing into the distribution that would
allow you to realize if cooling procedures or whatever played
a determining role in an *individual* case.
Best, BR
Cheers,
Roger Rowlett
Edward A. Berry wrote:
> mjvanraaij wrote:
>> how about this for a general idea:
>>
>> resolution is related to order in the crystals, more order = diffraction
>> to higher resolution
>> the order is determined by crystal contacts, stronger crystal contacts =
>> more order
>> more solvent means, on average, less close packing and less crystal
>> contacts
>> (of course, there are cases where only a few strong crystal contacts
>> make for a very highly ordered crystal diffracting to high resolution,
>> so there are quite a few exceptions of high solvent content being
>> compatible with high resolution diffraction)
>>
>> or is this too simplistic?
>> (and widely known?)
>
> Sounds very reasonable to me. One the positive side (and this may be what
> Roger was thinking of), for a given amount of protein in the
> asymmetric unit,
> increased solvent content means a larger unit cell and finer sampling
> of the
> molecular transform, -> more reflections and thus a greater data/param
> ratio
> at a given resolution. Also solvent flattening becomes more powerful.
> Ed
|