2,684
edits
Xds maxcc12: Difference between revisions
no edit summary
No edit summary |
No edit summary |
||
| Line 53: | Line 53: | ||
[[File:xds_maxcc12.png]] | [[File:xds_maxcc12.png]] | ||
The plot is useful because it shows you the cumulative influence of | The plot is useful because it shows you the cumulative influence of frames of the dataset on CC<sub>1/2</sub> and completeness of ten resolution shells (to change that number, you must modify the script). The highest resolution shell us usually the lowest curve (red); the curves above are lower resolution shells. (To see the legend which maps the colors and linetypes to resolution range numbers, remove the "set nokey" line in the script) | ||
This may shed light on the usefulness of certain frame ranges of your dataset which have high R<sub>meas</sub>. Do they really compromise CC<sub>1/2</sub> of the merged data - which is all you should care about? | This may shed light on the usefulness of certain frame ranges of your dataset which have high R<sub>meas</sub>. Do they really compromise CC<sub>1/2</sub> of the merged data - which is all you should care about? | ||
The example plot shows that CC<sub>1/2</sub> is highest around frame 60 to 70 and then gets lower due to radiation damage. However it also makes clear that around frame 60, the completeness is only about 50%. In this case, the anomalous signal is practically just noise. | The example plot shows that CC<sub>1/2</sub> is highest around frame 60 to 70 and then gets lower due to radiation damage. However it also makes clear that around frame 60, the completeness is only about 50%. In this case, the anomalous signal is practically just noise. | ||
Clearly, to reliably calculate CC<sub>1/2</sub> requires some multiplicity which is normally not available if the completeness is low. So expect a very noisy CC<sub>1/2</sub> plot at low completeness. At reasonable completeness, however, the plots are quite stable and you can nicely see what e.g radiation does: it hurts the high resolution shells and lets their CC<sub>1/2</sub> degrade. | |||
So the program may serve the purpose of helping to define the cutoff point beyond which frames are discarded due to radiation damage. I find that this cutoff point can be found satisfactorily with the help of this program. Of course the frame cutoff depends on the high resolution cutoff ! The procedure I suggest is: pick the highest resolution cutoff that has still significant signal (marked with "*" in CORRECT.LP), and define the frame cutoff as the frame where the CC<sub>1/2</sub> curve of this resolution range does no longer rise (i.e. becomes constant). | |||
Alternatively, you may base your decision on the anomalous CC<sub>1/2</sub> (bottom plot); the outcome may of course be different then. | |||