Would be great to have an equivalent to Dash Bio for Cheminformatics. I’d be interested in contributing.
2 Likes
Definitely! What components would be helpful?
1 Like
Personlay I am interested in data representation and visualization w.r.t. mass spectrometry.
Mass Spectrum (centroid) A typical diagram is a mass spectrum. To begin with the centroid representation would be easier than a profile representation. The diagram would be a variant of the Dash Bio Needle Plot. With the intensity as y axis and m/z as x axis. The range slider would be helpful as well. However it should be optional to allow the usual zooming in the diagram as well. In addition the y axis scaling should be selectable e.g. with a dropdown (no scaling, log10, etc.). Enabling data via selectors (“missense”) would be nice as is. Diagram highlighing like possible (“P53_TAD”) would be nice as well. However it would be required that a selector could map to several x-axis ranges in addition to a single x-axis range. Here the live Dash Bio Needle Plot for reference.
Seconding the request for a Dash Cheminformatics suite, one of the things that would be useful is some integration with common cheminformatics libraries. The RDKit is open source and widely used. The most widely used commercial package is OpenEye Toolkits.
We use Spotfire and the PerkinElmer cheminformatics tools as add-ins to visualize data.
The Dash for Drug Discovery example (https://dash-gallery.plotly.host/dash-drug-discovery/) is a nice example of the sorts of ways we visualize data. It would be advantageous, however, if we could show chemical structure when hovering over data points. Toolkits like the RDKit allow us to produce images which could be used in those hovers.
The molecule 2D viewer (https://dash-gallery.plotly.host/dash-molecule-2d-viewer/) is definitely a cool idea, but it doesn’t depict chemical structures the way chemists like. Chemists are most accustomed to (and prefer) ChemDraw. If forced, they will use the Marvin tools (from ChemAxon), but grudgingly.
In terms of 3D molecular visualization, I’ve been tinkering with https://3dmol.csb.pitt.edu/doc/index.html in Jupyter Lab, and I really quite like it.
I’d be happy to continue the conversation any time!
-Joann