openforcefield
Add visualization of parent hierarchy [low priority]
Add an optional visualization for the parent hierarchy of atom types. Parent dictionary should be able to be used to recreate this tree. Here is a conda installable module that might be useful: http://etetoolkit.org/docs/2.3/tutorial/tutorial_drawing.html
This is a low priority for now, but will be helpful in the future.
For the record, it may be helpful soon to have this along with the "score versus time" plots in #78 in order to understand why the score is changing when it does, at least for selected transitions. Clearly once we move to the ZINC parm@frosst set it's going to be very tricky to extract any useful insight from just browsing the log file, so being able to visualize the tree at points throughout the trajectory may be helpful.
I have this as a todo on my list right now, because I think we would like to have nice figures of these for the paper. I'll likely add whatever I come up with to the score_utils methods
I really like the style of the color wheel/tree just below the header "Node backgrounds"; not sure if it would work here, but I'm imagining something where the wheel is colored by element and different legs are labeled with their SMARTS/SMIRKS could be really cool if it could be made to work. Possibly too much trouble than it's worth to try and go with that layout (as opposed to a simple tree) though...
I'm working on this currently, I'm not sure how easy it will be to automate this process, but I have an ipython notebook generally showing how to do this in branch issue46 which has a work in progress pull request #214
I have some examples from this that I should put in utilities so other people can use it. I'll add cleaning up this notebook to my to do list and add it here.
The conda-installable networkx package can make circular tree plots, though I also came across a really nice example for making circular trees with d3.js using Python.
That is nice too, I used ete2 (linked above) to make the color coded circular plot David had in his ACS talk