BIOXHIT/CCP4: Ms 5.2.6 Extending the visualiser functionality

This document is a place to put ideas about extending the BIOXHIT/CCP4 visualiser functionality.

Suggested new functions

• Additional panes to display extra information: add new "panes" to show the list of jobs (similar to that in the current CCP4i), information about the files associated with a particular job and so on. Additional information.
• Remember setup between sessions: for example, the viewer could remember the last project that the user was looking at.
• Show the names of files on the linking lines: or have some way of accessing the files via text or an icon associated with a line that connects two job nodes.
• Display associated attributes: for example, show an icon that indicates that a job has an associated notebook entry. Clicking on the icon could allow the attributes to be manipulated (e.g. clicking on the notebook icon would open a window for viewing or editing the notebook entry for that job) (Charles Ballard).
  Another useful attribute to show would be if the job has any data harvesting files associated with it.
• Provide data-centric view: instead of the current view within the visualiser (which is a "process-centric" view that shows the steps taken joined by data flow), implement an alternative view that shows the data files as nodes joined by lines that represent the jobs (Jim Naismith/Rob Esnouf).
• Preferences options: provide a way for the user to customise the viewer, for example by setting the default orientation or the colours used for the different job statuses. Other possibilities:
  • Choice of "rendering engine" (currently defaults to dot, but could be e.g. circo)
  • Choice of file format for outputting the views to file e.g. gif, jpeg etc
  • Choice of font style and size (currently fixed to be the "regular" CCP4i font).
• Offer alternative graph layouts: by default the dot program is used to render the layout of graphs, however it might be useful to provide alternative layouts using the other Graphviz programs such as circo.
• Indicate when project is read-only: provide some indicator to the user when a project is read-only (for example, some text in the message bar and/or an icon of a padlock, or whatever).
• Colour-blind users: the visualiser currently uses green to represent FINISHED jobs and red for FAILED jobs, however this can cause problems for users with red-green colour-blindness. Possibilities include: using different colours, or colours with differing hues; using other visual indicators as well as/instead of colours to indicate status (e.g. a "tick" or a "cross" icon); allowing users to set their own preferred colours.
• Display "subjobs" from tracking database: allow expansion of a job node to display the substeps within that job, or else shrink the view of multiple subjobs back into one job (Serge Cohen).
  This requires the implementation of substeps or subjobs in the tracking database.
• Option to "View Files from Job" and "Rerun" jobs: add an option to view the input and output files associated with a job, and to rerun a job (cf the functionality in CCP4i). (Ronan Keegan).
• Summarise logfiles in bubblehelp for jobs: the bubblehelp associated with each job could include a brief automatically-generated summary of the job, for example by showing the initial and final Rfree values from a Refmac run (Ronan Keegan).
• Edit history data: enable the user to manually create or remove links between jobs.
• Search/selection tools: addtional tools could be added to help navigate the view, for example: selection tools could restrict the view to a range of job numbers or dates (e.g. "show only jobs run on 16th April 2007")
• Deposition functionality: allow the user to specify jobs as particular steps in the structure solution process, to help with deposition. One way to do this would be to allow the user to mark particular jobs in the view as the final refinement step, the data reduction step and so on. The viewer would then be able to collect harvesting and other data files, run checking programs to check for consistency, and pass the information to starKey to be written into the output XML.
  This would extend beyond the the viewer, but we could for example create a processor that could turn the starKey XML into a command for running PDBEXTRACT or collect the correct files for deposition.
• Expert system functionality: an "expert system" helper program could use the connections between the jobs to try and suggest what tasks could usefully be performed next (Chris Morris).
  I think that while this would use the "history" functionality, it is really a complementary application
• Add a key or legend to the display: this would be a small key that would explain what the colours of the boxes meant etc (similar to the key on a graph explaining which line is which).
• Enlarge or reduce nodes depending on "significance": nodes that represent more "significant" jobs (for example, REFMAC or MOLREP jobs) could be shown as larger than those with less significance (for example analysis or conversion jobs). Conversely, the less significant jobs could be shown as smaller nodes.
• Show inputs/outputs from jobs with no parents/children: for jobs that have no children, show nodes that represent the output files for that job (and similarly for jobs with no parents, show the input files).
• Annotation of jobs: allow the user to easily add a descriptive note to a job.
• Enable launching of external programs to view files: it would be useful to be able to launch Coot or CCP4mg directly from the visualiser in order to be able to view files from the jobs.

Additional information on suggested new panes

The figure below shows a simple schematic illustrating how extra "panes" could be added to the existing viewer in order to add new functionality. These are just suggestions and will need further thought.

The mock-up shows two new panes (labelled "1" and "2") added to the right of the existing graph view:

#1: Job list

This would be similar to the job list in the current CCP4i, namely a list of all the jobs currently shown in the display. The two views could be linked so that selecting jobs in one view (e.g. the graph view) would highlight the same jobs in the other view (the list view).

For both views things like "Delete/Archive" and "Rerun job" should be accessible from a menu that would appear on a right-hand mouse click).

#2: Job view

This would give a detailed view of the currently selected job (if only one job is selected). This should include: job number, title, taskname, driver application and file lists. The file lists should be "live" i.e. so that clicking on a file name would display the file.

This pane could be used for something else if there is more than one job selected.

Issues to investigate & resolve

• We should get feedback urgently from a wider range of non-programmer users (Martin Noble)
• Need to check that the relationships are preserved even if files in the system are overwritten or moved (Randy Read)

[Index] [Deliverables]

This file last modified