Ticket #386 (closed wish: fixed)

Opened 5 years ago

Last modified 5 years ago

spremembe v Feature Selection (bioinformatics)

Reported by: blaz Owned by: ales
Milestone: 2.6 Component: canvas
Severity: immediate Keywords:
Cc: Blocking:
Blocked By:


1) When cursor is over (or close to) any of the bars that change the selection of the histogram, let it change shape to <=>. It would be ok if user can move these barse only be left clicking on this bar (now, with a right click, you control the right bar, but this is a rather non-standard use of the right-click so change this to left-click based dragging of the bar). This is, left click on empty space does not do anything, you have to click on the bar and drag it to change its position.

2) Change the color of the selected area to match that of the histogram curve (blue)

3) Separate the commit part from the selection part (use the Commit box)

4) In Selection box, keep what is currently inside, but also add:

above the current contents a box "Threshold" and inside the box the two entry boxes where user can directly type in the values of the thresholds, or, when the user changes the threshold bars, the value in these boxes is updated accordingly (while dragging). If the score permits only one threshold, than the second box is hidden.

  • check box "Compute null distribution" (default: not checked)
  • label "Repetitions:" followed by entry box (just like for "n:", default 10)
  • button "Select w.r.t. null distribution"
  • label "p-value:" followed by the entry box (just like you did with GO widget), default is 0.01, but could go to 0.000001

If compute null distribution is not checked, everything below (from repetitions to the p-value box) is disabled.

  1. Some tests are only aplicable if a specific number of classes are present. E.g., can't compute t-test with three classes. The tests should thus be disabled/enabled according to the number of classes used.
  • number of classes = 2

disable anova, anova p-value

  • number of classes > 2

disable t-test, t-test p-value, (I do not know how you compute fold change, but probably need to remove this one as well)

  1. Reorder scores such that the order is:

fold change, log-2 fold change, t-test, t-test p-value, anova, anova p-value, signal to noise ratio, information gain, chi-square

Change History

comment:1 Changed 5 years ago by ales

  • Status changed from new to closed
  • Resolution set to fixed
Note: See TracTickets for help on using tickets.