Based on my previous posting question:
> I was hoping I could get some clarification on the difference between
> Rsym and Rmerge.
>
> Does the Rsym represent the differences in the symmetry related
> reflections on a single image?
>
> or,
>
> Does it represent the differences in the symmetry related reflections
on
> a single crystal?
>
> If it is the later, what about low and high resolution data collections?
>
> Do you report an Rmerge because you are comparing 2 different data
sets?
The general consensus seems to be that Rmerge is between data sets (only
from different crystals?).
It is still not clear to me when you collect 2 datasets (with differing
parameters, for example high
and low resolution) on the same crystal if you should report a Rmerge
or Rsym. From the responses
it seems the general standard is that you would still report it as
an Rsym.
However, it was pointed out that Rsym is reported for symmetry related
reflections on
a single image in MOSFLM.
Below is a listing of the responses I have received:
My favorite reply is from Jim Pflugraph,
Since you didn't get an immediate answer, here's a stab at it:
Rsym and Rmerge are often used interchangeably. But sometimes
they
are not. You need to check the documentation of the particular
program that is giving you numbers or the definition in the paper
you are reading or what the person you are talking to defines it/them
as.
Sometimes Rsym is within an image (i.e. MOSFLM Rsym) and sometimes from
reflns within a crystal. Rmerge usually includes these definitions
of
Rsym plus any other sources of reflns.
Manfred Weiss stated,
historically R(sym) stems from merging symmetry-related
reflections when data collection was still done by the
precession method. In that case, each reflection was only
measured one.
R(merge) was then used to describe the difference between
two data sets or between the parts of the data sets from
different crystals.
Nowadays, with the rotation method being the dominant
means for data collection, the distinction has become
meaningless.
What is often not considered, however, is that R(merge)
is dependent on the redundancy of the data. The higher
the redundancy the higher R(merge) even though the
averaged intensity should become more precise. You can read
a short description of that in J. Appl. Cryst. 30, 203-205
and in J. Appl. Cryst. 34, 130-135. In other words,
the quantity R(merge) is pretty much useless unless
the redundancy is also reported. Or else, a quantity called
redundancy-independent merging R-factor should be
calculated instead.
Other responses include,
Phil Evans: As far as I know, the terms Rsym & Rmerge are interchangeable
I am glad to see I am not the only one,
David Buckler: I appreciate your question and thought the basic differences
were along
the lines as you suggest in your question: Rsym is difference
between
multiple sym related spots (either same or multiple image) but Rmrg
was
for combining datasets -- and that different people seem to use the
terms in different ways. Now Phil Evans's reply makes me more
confused. Please either forward your summary of replies to the
whole
mailing list, or at least forward to me the individual replies if they
might help me, as you are seeking, to reach a better understanding.
Allan Mayte: My understanding has been that Rmerge is between
crystals, and Rsym for
symmetry mates within a single dataset (one xtal). If you conclude
something different from other replies, please let me know.
Marieke Thayer: Rsym is on a single crystal. Rmerge is between different
crystals.
Before data was consistently collected at cryogenic temperatures,
several crystals were needed to collect a complete data set.
The
Rmerge is usually higher because there are differences between crystals
that affect the relative intensities.
-- <> Paula Lario Nothing is ours, but time. |\ _,,,---,,_ ZZZzz /,`.-'`' -. ;-;;,_ _ |,4- ) )-,_. ,\ ( `'-' ,(_`> '---''(_/--' `-'\_)