Clicking on a point selects multiple points

cellprofiler-analyst

#1

CP Team,

I have two other problems now. One is a problem with selecting points in Scatterplots and DensityPlots. This does not happen with the example database, only when I use data from my pipeline. When I click on a single point in the scatterplot, it selects multiple points. Clicking on a different point also selects multiple points, just different ones for each point. I have attached an image of this:


I hope the other problem is simpler. When I display my images through CPAnalyst, the colors are off: the red and blue seem to be switched. In addition, I have 4 channels but only Red, Green and Blue are displayed. Is that something that the program can manage?

I am attaching the pipeline used to generate this data.

Thanks again for any help.
20090224 TissueFAX 4Chan DB PIPE.mat (2.07 KB)


#2

On the second point, you can go into File -> Edit Current Properties and ensure that “Pathway to Red/Green/Blue” images are correct. Be sure to explicitly Save Properties after you do this, since it won’t necessarily save them after you close the dialog.

I don’t know of any way to display or handle >3 channels in CPA 1.0. In the new CPA (2.0) this is planned to be functional, but the visualization tools are still in development.

As for the first point re: the multiple points being selected, this does seem to be a problem and I suspect that all objects from a particular image are getting selected. I will have to push this one over to our CPA programmer.

David


#3

Thanks, I was able to change the properties file and that took care of the colors problem.


#4

Hello,

The answer to your first question about the selection mechanism in scatterplot is that the scatterplot wasn’t really built for viewing per-object data because it is usually too large (in our experience) to display all of the points. What* is* happening when you select a point in a per-object scatterplot is CPA is looking up all other objects that come from that same image and selecting them as well. Unfortunately, most of the other scatterplot functionality either doesn’t apply to per-object data or is broken because it is untested on the data.

That said, you should be able to look at per-object distributions with the scatterplot (provided your data stays small), but you won’t be able to do other nice things like look at random images from your selection.

You could try using the density plot instead. This plot was designed with per-object data in mind.

Sorry things are a bit cludgy right now. CPA has fallen out of maintenance since we’ve been working on Classifier 2. On the plus side, we will soon be releasing a new version of Classifier 2 and begin work on a new set of visual analytics tools to complement it.

Best,
Adam