Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Nov 19 02:58
    cmdcolin commented #1556
  • Nov 19 02:56
    cmdcolin commented #1556
  • Nov 19 02:55

    cmdcolin on uniqueID_remove_gff3tabix

    Remove uniqueID from featureData (compare)

  • Nov 18 06:49
    Jungal10 commented #1556
  • Nov 17 20:36
    cmdcolin commented #1556
  • Nov 17 20:33
    cmdcolin commented #1556
  • Nov 13 19:04
    jrobinso opened #118
  • Nov 13 18:22
    Jungal10 commented #1556
  • Nov 13 18:21
    Jungal10 commented #1556
  • Nov 13 15:41
    cmdcolin commented #1556
  • Nov 13 06:24
    Jungal10 commented #1556
  • Nov 12 22:59
    cmdcolin commented #1556
  • Nov 12 22:39
    Jungal10 edited #1556
  • Nov 12 22:38
    Jungal10 opened #1556
  • Nov 12 21:52
    cmdcolin commented #1555
  • Nov 12 21:47
    cmdcolin commented #1555
  • Nov 12 21:46
    cmdcolin commented #1555
  • Nov 12 17:43
    GFJHogue opened #1555
  • Nov 10 00:38
    jrobinso commented #79
  • Nov 09 22:57
    cmdcolin commented #67
Scott Cain
@scottcain
@gringer yeah, I agree with Colin that server side is unlikely the problem (though I think there might be an apache module that could cache stuff, but I think generally people don’t have that turned on). I would double check that caching is turned off in your browser and that the track configs are really looking at the files you think it is. For example, I would check the in browser track config (“edit config” from the track label’s down triangle menu item), and then use curl -I to check the headers returned from the web server to verify that the modified date matches up with what you think it should be.
David Eccles (gringer)
@gringer
My attempts at generating minimal examples don't show the same issues, which is frustrating.
Changing the "urlTemplate" line to that of a different track also produces the expected results (i.e. the results of that other track).
From which I guess I conclude that the bigwig file is not correct?
Scott Cain
@scottcain
That’s what I was going to say—you could use command line tools to query the bigwig to make sure it has the numbers you expect.
David Eccles (gringer)
@gringer
... but this is a problem that's happening with the BAM files as well
Colin Diesh
@cmdcolin
one random thing to consider is that different depth programs can vary in slightly different ways
David Eccles (gringer)
@gringer
$ samtools depth raw/mm2_GC_0013_S2_called_BC12_vs_MmusG.bam -r chrM:4000-5100 | cut -f 3 | sort -rn | head -n 1
462
Scott Cain
@scottcain
@cmdcolin who’s responsible for covid19.jbrowse.org? It appears to be down.
Colin Diesh
@cmdcolin
Jbrowse filters out pcr dup, secondary and supp alignments by default
David Eccles (gringer)
@gringer

Here it is with all BAM filters turned off, no change in the 4-5.1k region:

max coverage still 8

David Eccles (gringer)
@gringer
Copied the BAM file to another file name in the same directory (test.bam), and it's working...
BAM track showing much greater coverage
David Eccles (gringer)
@gringer
But the md5 hashes for the files match:
[tracks.bw-GC_0013_S2_B16WTcDNA_BC12-both-track ]
storeClass     = StrandedPlotPlugin/Store/SeqFeature/StrandedBigWig
urlTemplate    = bw/mm2_called_GC_0013_S2_B16WTcDNA_BC12_vs_MmusG.bw
category       = MinION / Coverage
type           = StrandedPlotPlugin/View/Track/Wiggle/StrandedXYPlot
key            = Minimap2 coverage from B16.WT#A [GCOct20]
scale          = log
scoreType      = maxScore
autoscale      = global
style.pos_color = #228B22
style.neg_color = lightskyblue

[tracks.bw-test-track ]
storeClass     = StrandedPlotPlugin/Store/SeqFeature/StrandedBigWig
urlTemplate    = bw/dummy.bw
category       = MinION / Coverage
type           = StrandedPlotPlugin/View/Track/Wiggle/StrandedXYPlot
key            = Dummy coverage from B16.WT#A [GCOct20]
scale          = log
scoreType      = maxScore
autoscale      = global  
style.pos_color = #228B22
style.neg_color = lightskyblue
$ md5sum bw/dummy.bw.plus bw/mm2_GC_0013_S2_called_BC12_vs_MmusG.bw.plus
6cb081b194ba930818fd7099fe9bc81f  bw/mm2_GC_0013_S2_called_BC12_vs_MmusG.bw.plus
6cb081b194ba930818fd7099fe9bc81f  bw/dummy.bw.plus
Colin Diesh
@cmdcolin
thats pretty weird
@scottcain i hit the reboot button on covid19.jbrowse.org
David Eccles (gringer)
@gringer
I've renamed the copied files to "_allReads" and deleted the originals. I just hope I don't have to do this every time I update a track.
Colin Diesh
@cmdcolin
the most likely thing is client side cache
but i know you mentioned the other browser weirdness so hard to say
there is also a thing called cacheBusting where you can add a random string to the end of the url
but we dont do that generally
(i gotta watch movie now but lemme know if anything comes up)
David Eccles (gringer)
@gringer
Seems to just be those two tracks; other tracks where I've updated the underlying files aren't displaying the same issues
Scott Cain
@scottcain
@cmdcolin from Terence at NCBI: "Thanks again for the report. We’ve made some changes to our web server so that the “Content-Type: application/x-bam” header should be consistently returned. It looks like this has fixed the problem for IGV. Does that fix the JBrowse problem as well?"
Colin Diesh
@cmdcolin
should work now ya
Colin Diesh
@cmdcolin
i will check shortly but quickly looking at a previous example where it failed it is no longer returning content-encoding:gzip which is what we wanted
send terence a big thanks :]
Scott Cain
@scottcain
I sent him a response thanking him and his team and said it almost certainly fixed the problem but I would let him know if we run into anything else.
Scott Cain
@scottcain
@cmdcolin I think the wigglehighlighter plugin may have stopped working recently. At https://staging.wormbase.org/tools/genome/jbrowse-simple/?data=data%2Fc_elegans_PRJNA13758&loc=II%3A1..1758000&tracks=ALY-2%20Combined%20(GFP%20ChIP)_%20ALY-2_GFP_L2%20w%20peaks I get an error message like TypeError: s.config.highlightColor is not a function TypeError: s.config.highlightColor is not a function Is that more likely something I did, or something that happened in JBrowse core?
The bigwig and bigbed files both continue to exist. The bigbed has Content-Encoding: bzip2 and Content-Type: binary/octet-stream
Colin Diesh
@cmdcolin
might have changed a little bit
there was a user coding and iterating the wigglehighlighter codebase
lemme check
Scott Cain
@scottcain
Oh, and in case it matters, I’m using the dev branch for jbrowse.
Colin Diesh
@cmdcolin
could try to pull from wigglehighlighter master
it has an update
Scott Cain
@scottcain
I’m guessing I can supress that somehow
Colin Diesh
@cmdcolin
I think you have showLabels: true on there
or actually
can set showLabels:false
i guess its true by default
i could switch it if it seems easier
Scott Cain
@scottcain
Interesting—I don’t think the labels were there before. Anyway, making it false by default would be a lot easier since I have quite a few tracks, and I suspect most people would want it off in a context like this anyway.
Colin Diesh
@cmdcolin
ok
ill just switch it off
Scott Cain
@scottcain
👍
Colin Diesh
@cmdcolin
should be updated again
Scott Cain
@scottcain
Looks good; thanks @cmdcolin