IGV 3.0 bata with track hub does nothing obvious

[diekhans]

1. Install IGV 3.0 beta as over 2026-09-25
2. Track Hub -> Add Hub from URL
3. add https://public.gi.ucsc.edu/hausslerlab/t2t-h9-hub/hub.txt
4. No error and nothing has obviously been added

[jrobinso]

Hi Mark, I think we need to make some improvements WRT user feedback, but when I tried the hub was added to the Track Hubs menu. That is what is supposed to happen. However, it can't find any tracks to load (second screenshot). <img width="281" height="313" alt="Image" src="https://github.com/user-attachments/assets/f53b81ec-cfc3-4fbe-a256-3a5828fefaab" /> <img width="442" height="132" alt="Image" src="https://github.com/user-attachments/assets/6d154540-37f7-4aa8-a1aa-bf8f0c36a321" />

[jrobinso]

Oh never mind, I had hg19 selected when I loaded that hub, and it is correct to say there are no tracks to load for hg19. It showed up for hg19 because a genomes file defines a trackDb file for that assembly, but there are no tracks to load in it. After switching to hs1 the track hub worked.

[jrobinso]

Leave the issue open as a reminder to give more feedback. What would you expect to happen after loading a hub by URL? I think it should add it to the menu, AND open it, the second part is missing. You have to go find and select it. I can see how that would leave the impression that nothing happened.

[diekhans]

Ah, hg19 fooled me. However, the real issues that the new genomes, which are at UCSC GenArk, are not added. # t2t_h9_v01_hap1 # human (H9_T2Thap1 2026) - Homo sapiens (GCA_054883195.1) genome GCA_054883195.1 trackDb t2t_h9_v01_hap1/trackDb.txt groups groups.txt # t2t_h9_v01_hap2 # human (H9_T2Thap2 2026) - Homo sapiens (GCA_054883265.1) genome GCA_054883265.1 trackDb t2t_h9_v01_hap2/trackDb.txt groups groups.txt <img width="277" height="428" alt="Image" src="https://github.com/user-attachments/assets/cd7c56f2-cd8f-48d3-a8ba-c3392193cd1a" />

[diekhans]

I could seem to edit the above and add this image. I expected the GenArk genomes to appear her or to get an error

[jrobinso]

We might have different expectations for what loading a hub in IGV doese. Did you try to load those genomes through the Genomes menu? I found it quickly (see screenshots) , then the hub worked for loading tracks.

When you loaded the hub what was your expectation? It's been a while since that was implemented, so I'll have to investigate, but the purpose of hubs in IGV are for loading tracks into already defined genomes, where "defined" means genomes selected from the Genomes menu. We could look into changing this, of course.

[jrobinso]

The general flow in IGV is (1) define reference genome(s) by either selecting in "Load Genome From Host...", or from a local file or url, (2) load tracks from file, url, or defined hubs for that genome. This might differ somewhat from the UCSC flow. The hubs available under the Track Hubs menu will depend on the selected host genome from step 1.

I'll try inferring expcations from the thread above. When you loaded the hub your expecation was the listed genomes would appear in the pull down. Is that correct? That's not unreasonable, but might be difficult and would cause issues for some hubs that have a large number of genomes in "genomes.txt".

[maximilianh]

When I load a track or assembly hub, I expect the tracks to just appear in front of me. Also, my goal is to have a link from UCSC “open in IGV Desktop” for all track hubs, where the hubs are shown.

…

On Sat, Sep 26, 2026 at 13:13 Jim Robinson ***@***.***> wrote: *jrobinso* left a comment (igvteam/igv#1879) <#1879 (comment)> The general flow in IGV is (1) define reference genome(s) by either selecting in "Load Genome From Host...", or from a local file or url, (2) load tracks from file, url, or defined hubs for that genome. This might differ somewhat from the UCSC flow. The hubs available under the Track Hubs menu will depend on the selected host genome from step 1. — Reply to this email directly, view it on GitHub <#1879?email_source=notifications&email_token=AACL4TL263DICM7R3X5PX6D5Q72MTA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOBUHAZDCMZVHEYKM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#issuecomment-5848213590>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACL4TJFUM37RFPW4KQQ62T5Q72MTAVCNFSNUABDKJSXA33TNF2G64TZHMZTIMJWGU3TQO2JONZXKZJ3GU2TQOJYGUYDQOBWUF3AE> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/AACL4TM42DG6ZRFOW3ONLZ35Q72MTA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOBUHAZDCMZVHEYKM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSVGM33PORSXEX3JN5ZQ> and Android <https://github.com/notifications/mobile/android/AACL4TIQBXN62EIPXE7PNQD5Q72MTA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOBUHAZDCMZVHEYKM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>. Download it today! You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[diekhans]

Hi Jim

My expectation was that loading a hub would also "connect" to
the assembly in the same way as the UCSC browser. In IGV, I was
expecting it to be added to assembly pull-down menu at the top
left.

The general flow in IGV is (1) define reference genome(s) by
either selecting in "Load Genome From Host...", or from a
local file or url, (2) load tracks from file, url, or defined
hubs for that genome.

Even not knowning the implementation, I can see how IGV got
here.

However, this result in more steps, and is confusing since
doens't guide the user. With multi-assembly hubs, connecting to
the hub is the way to find what is in it. This means either
looking it up by using the UCSC browser or grabbing the
groups.txt file. Somestime the names are not what is expected.

The T2T H9 diploid assemblies in GenArk, used by the hub, are:

H9_T2T.hap1 Jan. 2026 human (H9 T2T hap1 2026) (GCA_054883195.1)
H9_T2T.hap2 Jan. 2026 human (H9 T2T hap2 2026) (GCA_054883265.1)

It wasn't obvious at first, but these are in "From host" !!
and now it all works.

I will hold off trying my hub with 464 human assemblies ...

Also, I think a generalization selecting tracks is related to:

#1283
Add ability to hide tracks without removing them.

it would be great to have one way to turn on and off tracks, regardless
of how one got the tracks into IGV.

[diekhans]

Also, my goal is to have a link from UCSC “open in IGV
Desktop” for all track hubs, where the hubs are shown.

I would have and will use this frequently... That would include going to the current genome I am at UCSC browser.

[jrobinso]

Thanks Mark and Max for the detailed explanation. I think I have an idea now of what's expected. It won't be 1-1 with the genome browser as the models differ, and we can't radically change the mental model. But I think we can come closer. Response to some comments follow

When I load a track or assembly hub, I expect the tracks to just appear in front of me.

When you select a track hub from the Track Hubs menu a dialog is opened presenting all tracks that are loadable in IGV. You then select what you want to load. You can return to the dialog at any time of course. In UCSC those tracks appear at the area in the bottom of the screen, rather than in a dialog.

Currently adding a track hub by URL is a different operation, it simply adds that track hub to the list of selectable hubs for assemblies referenced in the hubs "genomes.txt". I think it should do that, but in addition open the track selection dialog for the hub for the currently loaded genome assembly. This will help with the "nothing happened" problem, and is probably the expected thing as well.

If you select an assembly hub it should behave just like selecting a GenArk assembly from the Genomes menu, because that is what that menu does. IGV switches to that assembly, it's added to the pulldown, and the dialog for selecting default tracks for the assembly is opened. You can return to this dialog anytime to add more tracks.

My expectation was that loading a hub would also "connect" to
the assembly in the same way as the UCSC browser. In IGV, I was
expecting it to be added to assembly pull-down menu at the top
left.

Got it. For hubs that reference UCSC assemblies IGV knows about we can do that. However there is a cost if there are multiple assemblies referenced in the hub. To add it to the pulldown we have to do an http query for some meta information (like the name, url to the twobit, cytoband if available, etc), and create a local genome json file for the assembly. If it's a GenArk assembly this involves a query for the hub.txt file, if not it queries IGV's genome json lists. The later includes hg38, hg19, mm10, etc, maybe 10-15 total.
If the ID isn't found in either of these lists there is not enough information in a hub for us to get the required information.

This is not a problem for a handful of assemblies, but 464 for example would cause at a minimum 464 http query round trips, and perhaps a much longer pulldown than the user wanted. Perhaps they just wanted a few tracks for one of the assemblies. So if the list is large maybe we should present a selection dialog with all referenced assembly IDs, and let the user choose.

Add ability to hide tracks without removing them.

There's no functional difference in IGV, but I think you mean hide tracks without having to right-click and "remove". We have that in igv-web, and I thought we had it in IGV as well but I just confirmed we don't yet. Will put that on my list. In igv-web the track checkboxes in the track selection dialog become toggles, unchecking a box will unload the track. I think that is what you are looking for, correct me if I'm wrong.

[jrobinso]

One point on hiding tracks, the above applies to track hub tracks. If a user loads a track by URL or from a local file there is no selection dialog to interact with. We could create one for this case, but I'm not sure it's worth it.

[diekhans]

I just made great use of IGV with assembly hub (hg38 + a plasmid) to understand why an analysis was producing results that made no sense!! Made two peoples day.

You can return to this dialog anytime to add more tracks.

Although it seems you can not remove them, They are grayed out and can not be unchecked. That was frustrating. I wanted to look at a whole different set of tracks without the others in the way. I can press remove on each track, but that is a lot more mouse clicks.

To add it to the pulldown we have to do an http query for some meta information (like the name, url to the twobit, cytoband if available, etc), and create a local genome json file for the assembly.

Is it necessary to get all information for all genomes upfront? Could it be done in a lazy manner when the assembly is actually accesses? Get just enough to the descriptions for each genome upfront, not the tracks.

This is not a problem for a handful of assemblies, but 464 for example would cause at a minimum 464 http query round trips, and perhaps a much longer pulldown than the user wanted.

That is an extreme case, not something to optimize for at this time. Just nice if it doesn't break. I do weird things, it will not be common.

> Add ability to hide tracks without removing them. There's no functional difference in IGV, but I think you mean hide tracks without having to right-click and "remove".

Yes, that is what I mean. I view an IGV session as something like a track hub builder where I add more tracks as I work and then, then save the session. If I want more vertical space, I want to hide the tracks and not have to remove and read URLs. I could a igv-session-to-trackhub utility.

[maximilianh]

I could a igv-session-to-trackhub utility.

~kent/src/utils/hubtools/hubtools import igv <filename>

…

On Tue, Sep 29, 2026 at 6:30 PM Mark Diekhans ***@***.***> wrote: *diekhans* left a comment (igvteam/igv#1879) <#1879 (comment)> I just made great use of IGV with assembly hub (hg38 + a plasmid) to understand why an analysis was producing results that made no sense!! Made two peoples day. > You can return to this dialog anytime to add more tracks. Although it seems you can not remove them, They are grayed out and can not be unchecked. That was frustrating. I wanted to look at a whole different set of tracks without the others in the way. I can press remove on each track, but that is a lot more mouse clicks. > To add it to the > pulldown we have to do an http query for some meta information > (like the name, url to the twobit, cytoband if available, > etc), and create a local genome json file for the assembly. Is it necessary to get all information for all genomes upfront? Could it be done in a lazy manner when the assembly is actually accesses? Get just enough to the descriptions for each genome upfront, not the tracks. > This is not a problem for a handful of assemblies, but 464 for > example would cause at a minimum 464 http query round trips, > and perhaps a much longer pulldown than the user wanted. That is an extreme case, not something to optimize for at this time. Just nice if it doesn't break. I do weird things, it will not be common. > > Add ability to hide tracks without removing them. > > There's no functional difference in IGV, but I think you mean > hide tracks without having to right-click and "remove". Yes, that is what I mean. I view an IGV session as something like a track hub builder where I add more tracks as I work and then, then save the session. If I want more vertical space, I want to hide the tracks and not have to remove and read URLs. I could a igv-session-to-trackhub utility. — Reply to this email directly, view it on GitHub <#1879?email_source=notifications&email_token=AACL4TNJCX6NLI47WP3TVQD5RRO4BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOJQGIZDQMJUGYYKM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-5902281460>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACL4TPLSEIWPQKWUALQUKD5RRO4BAVCNFSNUABDKJSXA33TNF2G64TZHMZTIMJWGU3TQO2JONZXKZJ3GU2TQOJYGUYDQOBWUF3AE> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/AACL4TL5QAWF65MV6U4LQ2D5RRO4BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOJQGIZDQMJUGYYKM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJKTGN5XXIZLSL5UW64Y> and Android <https://github.com/notifications/mobile/android/AACL4TLX5MXN6WCN7CIYPB35RRO4BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKOJQGIZDQMJUGYYKM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>. Download it today! You are receiving this because you commented.Message ID: ***@***.***>