NLDR plot axes have fixed default ranges

[jembrown]

NLDR plot axes seem to use fixed default ranges, regardless of the coordinates generated by an NLDR analysis. This requires panning and zooming to effectively view the plot.

Could axis ranges be set based on the NLDR coordinates for a particular analysis?

[reid-wagner]

Hi @jembrown,

Were you encountering this with 2D data?

3D plot ranges appear to fit the data comfortably, but I did find an issue with 2D plots where the visualization code would unnecessarily include the positive range of the same magnitude for negative coordinates, causing the points to be clumped up and offset from the center of the plot. For example a single point (-100, -20) would have ranges [-100, 100], [-20, 20].

I removed the custom ranges from 2d plots so that it's automatically ranged by plotly, as is done for the 3d plots, and included this change in the latest CloudForest image. Let me know if that helps!

[reid-wagner]

A couple of notes about trying the new code:

1. You may want to clear your browser cache, as the visualization code can be cached. Looks like this is a known caveat with Galaxy but I'd like to not have to do this.
2. As we've discussed, if you use an old volume with the new image you won't see the updates. I'm hoping the process described here can be automated to some degree.

[jembrown]

Thanks, @reid-wagner. Yes, I encountered this problem with 2D plots. Setting the range automatically based on the coordinates of the projection sounds like a good idea. One question - are the x- and y-axes set to have the same scale or are the scales set independently? I do worry some that the plots could become distorted (could always look very "round") if one axis is stretched relative to the other.

[reid-wagner]

@jembrown - the axes now are scaled independently, but the scales can be set to an equal ratio. I can add that as a drop down option, next to the Subset Plot drop-down. Which would you prefer the default to be?

[jembrown]

@reid-wagner Ok, thanks. Having an option would be great, but my sense is that the best default would be for the axes to have the same scale.

[reid-wagner]

Sounds good. The latest version has equal-scale axes. Note that I've changed the usage of the run.sh script so that instead of sh run.sh <tag> <container_name> [volume] it's now ./run.sh [-v volume] <tag> <container_name>.