DP (dot plot / pairing) format bug: arcs cross track boundaries

[scottkall]

I'm seeing a bug where large arcs from a DP (here converted to ".dp.bp") file (see IGV-supported RNA Secondary Structure Formats) cross Track, but not Panel, boundaries. My hack solution is to separate the tracks into separate, necessarily differently-named Panels in the XML session file. Confirm this is a bug? See screenshot examples attached. Could you fix this?

I'm running IGV 2.4.6 on MacOS 10.13.3.

Here are the first four lines of the underlying .dp file:

1816
i   j   -log10(Probability)
6   37  1.62994
7   36  1.59094

And here are the first six lines (chromosome name redacted) of the corresponding, IGV-auto-converted .dp.bp file:

color:  255 204 0
color:  72  143 205
color:  81  184 72
nameRedacted    46  47  1811    1812    0
nameRedacted    92  93  105 106 0
nameRedacted    95  95  100 100 0

Bug example:

My hack: separate Panels:

[jrobinso]

The only possible fix in the current architecture would be to truncate the arcs so they don't cross the boundary. You would not get a scroll bar or other means to see the truncated arcs. Is this what you are requesting?

[jrobinso]

In my forum response I suggested separate panels as you are doing because I thought that would give you scroll bars, but I see from your screenshot that it does not. That could also be considered a bug.

[jrobinso]

There are 2 potential fixes. (1) just fix the arc renderer so it doesn't draw into other track spaces. Arcs are truncated. No scroll bar. (2) automatically wrap arc tracks in new panels, as is done for bam files, and also potentially add scroll bars if arcs are truncated. Feedback on these?

[scottkall]

@jrobinso: My colleague working on this may comment himself further, but he said to me:

"I think solution 1 where the arcs are truncated and do not spill over into the next panel is fine. Most of the arcs are localized and a scientist familiar with this type of visualization would recognize arc going off the screen as sites of long-range interactions."

It sounds like solution 1 could be easier/faster to implement, too? If so, that would be great.

[jrobinso]

Yes, solution 1 is much easier. I'll update here when fixed.

On Fri, Jan 26, 2018 at 8:11 AM, Scott Kallgren notifications@github.com wrote:

@jrobinso https://github.com/jrobinso: My colleague working on this may comment himself further, but he said to me:

"I think solution 1 where the arcs are truncated and do not spill over into the next panel is fine. Most of the arcs are localized and a scientist familiar with this type of visualization would recognize arc going off the screen as sites of long-range interactions."

It sounds like solution 1 could be easier/faster to implement, too? If so, that would be great.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/igvteam/igv/issues/511#issuecomment-360827987, or mute the thread https://github.com/notifications/unsubscribe-auth/AA49HKdkX5J-8cy3q9uUfhO836MjQOoiks5tOfkVgaJpZM4RtIqY .

[jrobinso]

This should be fixed now in 2.4.7

[dmmauger]

@jrobinso : The 2.4.7 release appears to have a bug in the structure visualization update. If multiple base pair tracks are loaded only the top track is displayed. See example below.

In this example three different .bp files are loaded and only the top one is displayed. Removing the top track will allow visualization of the middle track.

MacOS 10.12.6 with IGV 2.4.7

[jrobinso]

OK, this is probably from setting the clip. I thought that was too easy. So use 2.4.6 for now until this is fixed, which shouldn't be long.

On Wed, Jan 31, 2018 at 11:56 AM, dmmauger notifications@github.com wrote:

@jrobinso https://github.com/jrobinso : The 2.4.7 release appears to have a bug in the structure visualization update. If multiple base pair tracks are loaded only the top track is displayed. See example below.

In this example three different .bp files are loaded and only the top one is displayed. Removing the top track will allow visualization of the middle track.

MacOS 10.12.6 with IGV 2.4.7

[image: example 1] https://user-images.githubusercontent.com/15740949/35644178-47163014-0696-11e8-9b9d-d55287bd785b.jpg

[image: example 2] https://user-images.githubusercontent.com/15740949/35644171-40f6fab0-0696-11e8-8402-6d037dde2df1.jpg

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/igvteam/igv/issues/511#issuecomment-362051491, or mute the thread https://github.com/notifications/unsubscribe-auth/AA49HFtbeLJWHb2bcAlklSs6hhEhQktJks5tQMVTgaJpZM4RtIqY .

[jrobinso]

I think this is fixed, could you try the snapshot build (link below). If the fix is confirmed I will release this as 2.4.8

http://software.broadinstitute.org/software/igv/download_snapshot

On Thu, Feb 1, 2018 at 12:54 PM, James Robinson <jrobinso@broadinstitute.org

wrote:

OK, this is probably from setting the clip. I thought that was too easy. So use 2.4.6 for now until this is fixed, which shouldn't be long.

On Wed, Jan 31, 2018 at 11:56 AM, dmmauger notifications@github.com wrote:

@jrobinso https://github.com/jrobinso : The 2.4.7 release appears to have a bug in the structure visualization update. If multiple base pair tracks are loaded only the top track is displayed. See example below.

In this example three different .bp files are loaded and only the top one is displayed. Removing the top track will allow visualization of the middle track.

MacOS 10.12.6 with IGV 2.4.7

[image: example 1] https://user-images.githubusercontent.com/15740949/35644178-47163014-0696-11e8-9b9d-d55287bd785b.jpg

[image: example 2] https://user-images.githubusercontent.com/15740949/35644171-40f6fab0-0696-11e8-8402-6d037dde2df1.jpg

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/igvteam/igv/issues/511#issuecomment-362051491, or mute the thread https://github.com/notifications/unsubscribe-auth/AA49HFtbeLJWHb2bcAlklSs6hhEhQktJks5tQMVTgaJpZM4RtIqY .

[dmmauger]

I can confirm that the bug is fixed in the snapshot build. Screenshot attached below.

Thanks!

[jrobinso]

OK, we'll release it as 2.4.8 this weekend.

BTW you can suppress the non-meaningful sample attributes shown there by selecting "View > Preferences" and unchecking "Show Attribute Display"