Open tasks for v1.0.0

[dominiquesydow]

core infrastructure

• Add Ligand class - see Update core classes w.r.t. dynophore 0.2 output #21
• Use superfeature colors as defined in dynophore-0.2 output instead of hard-coded feature colors in definitions; update plots/nglview accordingly - see Replace hardcoded feature colors with superfeature color attribute #22
• Use Ligand as Dynophore attribute? If ligand PDB block were part of JSON/PML files, that would be fantastic! Simplify rdkit molecule generation for ligand #31

Plots (static/interactive)

• Interactive plots: Too slow for 20k frames. Optimize.
• Plots: Heatmap: Add x/y labels
• Add slider for frame range (min/max)
• Add % to plot labels / legends
• Use matplotlib.pyplot.eventplot for occurrence plots instead of simple dot plots... - see Use plt.eventplot for superfeature/envpartner occurrences #23
• Interactive plots: Also select envpartners (depending on superfeature)? > Set minimum percentage cutoff to filter out low-coverage envpartners and superfeatures - see Use plt.eventplot for superfeature/envpartner occurrences #23
• [TBD] Distance plots: Add distance cutoffs or interaction on/off - see Distance plots: Show interaction-relevant distances #35
  • [Medium] Add marker for feature cutoffs (needs feature definitions added to pml file)
  • [Minor] Or add widget that toggles between "all frames" and "only frames where superfeature occurs"
  • Distance line plot: Add interaction-relevant distances as dots on-top
  • Distance histogram: Plot interaction-relevant distances only
• [Minor] Add tick to collapse envpartners with same residue/different atoms for one superfeature (e.g. ASP 89) - see Collapse envpartners with same residue and superfeature #36

3D viz

• Feature clouds by superfeature ID? At the moment, pml only contains H-123344123124 IDs - can we add H[1234, 1235, ...] to pml?
• Check if dcd file has appropriate number of frames (e.g. 1000 frames) > For 20k frames this was not a problem in the Jupyter Notebook
• Add interpolation to nglview trajectory > Added nglview.player.TrajectoryPlayer(view) to notebooks.
• Add frame index to points in PML?
• [Minor] Add pocket residue side chains: We do not have all pocket residues in PML/JSON output but we can add to each superfeature the environmental partners. Extend NGL dynophore visualization #28
• [Major] Temporal resolution of dynophore clouds (use point frame index) - see Color clouds by frame #32
  • Color clouds by frame index: What kind of color gradient shall we use? Keep feature color and use sequential colormap? Matplotlib has colormaps but can we use our hex colors as endpoint? HSV color model: https://scikit-image.org/docs/dev/auto_examples/color_exposure/plot_rgb_to_hsv.html
  • Select cloud points by frame index range: Add slider that allows to select cloud points for selected frame range only? Do we need this?

2D viz

• 2D viz: Add 2D ligand view with highlighted features, as done in KinFragLib - see Add ligand 2D view #17

CLI

• Make dcd optional!
• Add dynoviz demo

Package

• Make package conda-installable (help from Jaime)
• Make raise_keyerror_if_invalid_superfeature_name private!!

Documentation

• Add prominent link to tutorials?
• Fix code formatting here: https://dynophores.readthedocs.io/en/latest/installing.html > Decided it is ok to have some words highlighted in a different color.
• [Question] How to save widget images? (Missing on GH and RTD) Plots notebook: Save widget state #29

[nadja-mansurov]

Hi @dominiquesydow

I have a question regarding

Interactive plots: Also select envpartners (depending on superfeature)?

Is it for 2d view? What tool are you going to use?
It looks like, that d3.js library could be imported to the notebook. It might be useful to create an interactive visualization of interaction

However, if it is only about plots, Altair looks better

[dominiquesydow]

Hi @nadja-mansurov,

The interactive plots are already in-place using ipywidgets (code).
Interactive plots are e.g. used in this notebook.

Unfortunately they not displayed in the website-rendered version, to see them you would need to run the notebook locally.

Here a screenshot:

I have a question regarding

Interactive plots: Also select envpartners (depending on superfeature)?

The idea was to add another widget that allows to select the environmental partners (rows) but they depend on the selected superfeatures and that becomes complex.

Instead, I was wondering if we offer widget for a percentage cutoff ("show only all env partners that appear more often than X% of the frames").

What do you think?

[dominiquesydow]

It looks like, that d3.js library could be imported to the notebook. It might be useful to create an interactive visualization of interaction

The d3.js import option for notebooks is amazing!

Do you think this could help us to add interaction partners to this 2D ligand view?

What I have is

(from dev notebook)

What I would like to add around the ligand are the env partners somehow like here:

(from my master thesis by hand)

[nadja-mansurov]

@dominiquesydow thank you for the quick response!
I guess it is possible to create a visualization like the one from your master thesis (thanks a lot for the link! :) )

I can imagine, that 2D visualization will be a "background" and env partners will be added by d3.
Another way is to create a visualization from scratch using a JSON file and d3.js.

I think I will create a separate branch from view-2d, and as the first step just add d3.js there

[dominiquesydow]

All features for a version 1.0.0 are implemented - moved a few more technical issues to #25. Closing this issue.