click on a BAM read feature does not select it, selects a different read instead

[GreggHelt2]

Bug found by Moni:

050212
What: WebApollo
Position: Amel 4.5, Group15.9:294199..295667
Summary: the cursor is positioned where I am clicking to select an RNA-Seq read (from the "Forager RNA-Seq reads" track), BUT the read that gets selected is the one highlighted in red elsewhere in the picture.
Caveat: it may be 'location-specific' and 'zoom-level-specific', (or maybe I'm just a bit obsessive), so I am unsure of its relevance. It has happened more than once, even after refreshing, and logging out and back in.

The associated photos (located at ...) show almost the same thing: (Ed & Gregg: I Apollo-gize for the poor quality of the pics, but I don't know how to make screenshots without moving the cursor). :)

• WebApollo-troubleshooting-misplaced_cursor1.jpg: this is a close up, because the read that ends up being selected is near the cursor.
• WebApollo-troubleshooting-misplaced_cursor2.jpg and WebApollo-troubleshooting-misplaced_cursor3.jpg: in these two cases the track that gets selected is a bit farther away (cursor at top, selected read at bottom).
• WebApollo-troubleshooting-misplaced_cursor4.jpg: this is what it looks like when it is selected correctly.

Reproducible?:
I can consistently reproduce it with reads located towards the beginning of the column, on the right hand side, between tick marks 294,500 and 294,750 (see WebApollo-troubleshooting-misplaced_cursor5.jpg).

Other Notes:
Maybe I'm splitting hairs here: it happens at this resolution (Group15.9:294199..295667), and again if I click the zoom-in button once, but disappears if I zoom in closer, or out.

---

[GreggHelt2]

Need to add links to images from Moni that show issue

[GreggHelt2]

My best guess is that this is an issue of clashing IDs, where
WebApollo assumes that IDs are unique but there are actually two or
more features claiming the same ID. So you could click on one feature
but it might select/highlight a different feature with the same ID.

I did a few things to try and uniquify BAM data IDs, but I'm sure I
haven't seen all possible BAM quirks. I'll try and test this theory
today (okay really tonight) in that region youre looking at, and
hopefully figure out a fix.

Trying to debug this does point out an ability in JBrowse that's
lacking in WebApollo, which is to click on any feature and pop up an
info dialog showing start, stop, ID, etc. If we still had that
ability you could directly compare IDs to test out this theory. I've
already been thinking we should bring something like the info dialog
back, but we're already using click for selection. Maybe
click+modifier? But we're already using click modifiers for a few
different actions. Maybe a hover-over popup? But would need a delay
-- I hate sites where lots of popups are triggering when you're just
moving the pointer across the screen. Or maybe it's time to add a
right-click context menu to non-annotation features, with a "Show
Info..." option. Hmm...

[GreggHelt2]

Verified that problem is with multiple BAM features getting assigned the same ID.
IDs are currently assigned as READNAME + "/" + START + "-" + END.
Looking at the BAM records, for the case shown in WebApollo-troubleshooting-misplaced_cursor2.JPG, the assigned ID for both features is "HWUSI-EAS594-R_0059:2:100:6076:11113#TTAGGC/294595-294695". The features themselves are identical, or at least any differences are not carried over from the BAM records to the WebApollo JSON data structs. Not sure why there would be two identical entries in the same BAM file.