Short of sitting with someone and use Coot with them watching and vice-versa, I am wondering whether there is something more efficient like for example a model and map in which certain common problematic regions have been introduced on purpose.
I am not aware of a tutorial with predefined errors to learn model building.
I would suggest to practice your model building skills by re-determination of already known structures. First, download experimental structure factor amplitudes for a protein of known structure. A good place to download datasets at various resolutions is the Autostruct website (http://www.ccp4.ac.uk/autostruct/testdata/index.html). If you want to "simulate" the whole process of protein structure determination by MAD, SAD or SIRAS or MIR, this is the place to go. If you want to start from the very beginning, you can get raw diffration images at the TARDIS website (http://tardis.edu.au/).
After calculation of initial phases for you dataset, you can use various autobuilding programs (e.g. phenix, arpWarp, solve/resolve) to give you a good start. Continue by manually checking your structure and improve it with coot. You can also try to manually build the whole molecule or certain regions of your molecule in coot. Just have the published structure loaded in coot and switch in on from time to time to check your progress and identify building errors.
Hi,
if you don't want to go through the whole process of structure determination, you can go to http://www.embl-hamburg.de/xtutor/ . You will find several tutorials for structure determination using various phasing methods, but you can also find all the files generated in the process. You can therefore pick the phased reflection files and starting models...
- Badges
- Science method
- Similar topics
- Teaching