72 View 0 Comment New Comments enabled In the category: General Discussion

Please see the slides, diagrams for MRCM and information about them from Ed Cheetham.  We would like your input on how we can utilise these diagrams.

Ed says:

The attached MRCM-post-all.dot, MRCM-post-all.svg, MRCM-pre-all.dot and MRCM-pre-all.svg are the ‘dot’ files for the images generated from the MRCM Domain international refset templates, along with their corresponding svg output files, using GraphViz’s (https://www.graphviz.org/) dot.exe layout.

I’m sure you already know this, but essentially once you have installed the set of GraphViz executables, transformation from dot to svg is achieved by a command line instruction of the form: 

    dot -Tsvg /…path to dot file…/MRCM-pre-all.dot -o /…path to output file…/MRCM-pre-all.svg

    See https://graphviz.gitlab.io/_pages/doc/info/command.html for details.

Different layouts can be achieved with different executables in place of dot (neato, circo etc.) and different output formats by varying the -T… flag (-Tjpg, -Tpng etc.).

As discussed on the call dot files (or their text) can also be consumed directly by other applications. I have previously favoured ZGRViewer (http://zvtm.sourceforge.net/zgrviewer.html) which is a bit more interactive, although I suspect this is starting to suffer from lack of maintenance, and others mentioned the D3-graphviz/Viz.js project – thanks for this, I shall experiment with this myself. Graphviz has its own GVEdit.exe program, and Gephi (https://gephi.org/) is fun for a completely different experience. I was only able to import the dot file directly into an earlier version yesterday, however this sample file should load (apologies if it doesn’t) into the current 0.9.2 version and allow you to play with a number of different layouts, statistics etc.:

    MRCM-pre-all.gephi

A very fast and dirty approach is just to copy the text of a dot file and past it into the input window at http://www.webgraphviz.com/. The layout for the splines=ortho setting is a bit odd, but otherwise it seems to work nicely with a standard ‘dot’ executable.

In addition there are some smaller sample files:

    MRCM-pre-observable.dot – just the observables (pre-coordinated) dot file

    MRCM-pre-procedures.dot – just the procedures (pre-coordinated) dot file

    MRCM-pre-drugs.dot – a sample of ‘themed’ domains – administration of substance plus all the pharmaceutical/substance domains.

  • Regarding the files themselves, and as largely mentioned on the call:
  • The svg outputs (and possibly others) should have tooltips on each edge and model-participating class, and each class should have a URI/link which takes you to the international browser.
  • If you wish to work in other applications, or do more than just render the dot/svg output then the caption and key might be a nuisance. The relevant lines in each dot file (towards the end, just open it in a text editor – but not Word!) can just be removed, or commented out by wrapping any unwanted sections in /*…*/ comment markers.
  • You can experiment with different edge flows by editing the rankdir=LR instruction in the top line: LR indicates Left->Right, so you can replace this with BT, TB or RL (without taking mitigating steps BT and TB will cause the ‘caption’ to misbehave).
  • Removing splines=ortho; instruction will result in a rendering with curvy edges.


Have fun and thanks again.


MRCM presentation 20200213.pptx

diagrams.zip  includes following files: 

MRCM-post-all.dot
MRCM-post-all.svg
MRCM-pre-all.dot
MRCM-pre-all.gephi
MRCM-pre-all.svg
MRCM-pre-observable.dot
MRCM-pre-procedures.dot
MRCM-pre-drugs.dot


Contributors (0)

Comments

    Add new comment