Serena Cooper, Manchester University
I was persuaded to attend the recent CCP4 workshop at the IUCr meeting in Glasgow. As a CCP4 user I was sceptical that I would learn anything new, however I was pleasantly surprised, or perhaps it was a reflection on how little I already knew!
The workshop began with a general overview and update of CCP4 by Martyn Winn. Although most of it was not new to me there were a few things I was unaware of, for example, map files could be dumped by mapdump or that you can now assign project and dataset names to different datasets within MTZ files. Phil Evans talked about SCALA and the best route for data reduction from various image processing packages, and the optimum ways to scale your data. He also gave a list of what to check in the output, in particular he stressed the importance of the normal probability plots. These can be obtained by using the keywords NORMPLOT and ANOMPLOT and viewed in Xmgr, although there was some debate to whether it would be more appropriate to include them in loggraph. Eleanor Dodson talked through using the GUI (despite a few technical problems) and certainly tempted me to try using it.
In the morning we were polled on what people wanted to have more specialist information on. Five discussion groups were set up for the afternoon session, these being for Refmac, MIR strategies, Mosflm, SCALA and the GUI. There was only time to attend two of these groups so I can't comment on the MIR, GUI or SCALA sessions.
Garib Murshudov gave examples of how to run Refmac, in particular using NCS. He also stressed the importance of including hydrogens in anisotropic calculations. He suggested a good way to get initial anisotropic values for co-factors or metal centres is to use the aniso keyword, which will calculate aniso cards for all atoms if there aren't any and then you can extract the relevant atoms. What most people wanted to know was what to check in the log files, and he gave a list of what he looks at: R; Rfree; A estimations; B's ; rms deviations in bonds, angles torsion angles and chiral volumes, the latter two being good indicators of where the model is poorly fitted.
Harry Powell was answering questions about Mosflm. It quickly became apparent that there were a lot of people who had either never used Mosflm or not within the last few years and were unaware of the recent improvements (and that there was an alternative to Denzo). He therefore gave an outline of how to run it and what to do if you have weak images, advising that the program works best using 200-500 spots. The session then moved on to general questions and there was a bit of a Denzo versus Mosflm discussion. From the tests Harry has been doing Mosflm is comparable speed wise and is no less capable at indexing. Mosflm has the added advantages of being free, with most detector types easily available and the support is helpful.
There were also opportunities throughout the day to have discussions
with the programmers to sort out more individual queries. At the end of
the workshop I felt that I had learnt something, been reassured that what
I was doing was correct and that the programmers were approachable and
always willing to help.