On 28 February 2010 22:25, Michael Bieri <michael.bieri@xxxxxx> wrote:
How about opening a window that displays a list of spin systems/residues. The user can then select graphically residues to exclude (multi select object) ?
This is what was thinking about for the future. The problem now is that you don't have access to this list of spins yet. But for the future, the spins can be loaded prior to execution of the auto-analysis. I.e. a special pipe can be created by the GUI for this frame, and the sequence loading (from file or from PDB, both should be available). Then you use a TreeView widget to display the sequence data (see for example the TreeCtrl.py wx demo), where the first level is molecule, the second is residue, and the third is spin. With this tree view, maybe on the left hand side, then there could be actions for changing it's selection value. This could even change the Tree icon (a special picture we make for this) to grey to show it is deselected. For now though, I would suggest something simple. What do you think? In the end we can modify the API to accept a pipe name, and then the auto-analysis will copy all the data from this GUI specific pipe rather than using all the current arguments. This would be very powerful for the scripting UI as well, as then most of the features of relax for spin manipulation are available again. As a crazy future idea, I would suggest that one analysis type has many different 'views', maybe selectable with bitmap buttons on a side or top toolbar. I.e. spin mode with the TreeView where clicking on each element brings up some GUI element showing all the contents of the spin (or residue, or molecule), a structure view with pymol integrated, maybe a results view, etc.
We need to let the gui know what kind of molecule it is, but maybe in a later version. I suggest to release the gui 1.00 as a protein tool and then expand it to others. What do you think? This then could be communicated as a small article to point specifically that relax may deal with other macromolecules.
I don't think we need to make any restrictions for now. There are no restrictions in the core of relax, and I'm trying to help you in the design so that you don't specifically add anything which adds restrictions. This will not be hard. The "unresolved residues" part needs to be redesigned anyway, so why not design it correctly from the start? Things like specifying the heteronucleus and proton names should be up to the user too. PDB files of proteins are far too diverse for us to start guessing atom naming! The user needs to tell relax what weirdness is in that file so that the NH vectors can be properly extracted. Backbone nitrogens are not always called 'N' and the attached proton name is often 'HN' rather than 'H'. Cheers, Edward