No reads shown for MT variants in internal IGV

[Jakob37]

Describe the bug

Previously (before updating 4.81 -> 4.88), when navigating to an MT variant page, it was possible to bring up the IGV viewer track including coverage tracks.

Now coverage tracks are only present for non MT variants.

A working variant:

A non working MT-variant:

Accessed from this "IGV viewer" button:

Additional info

I saw the IGV JS was updated in 4.88.1, so I tried running on that version. Same situation there - no MT tracks in neither 4.88 nor 4.88.1

A recurrent issue is that we here in Lund call the mitochondrial chromosome "M" instead of "MT". Might it be something related to that?

[northwestwitch]

Hi @Jakob37, where are your MT variants read from, same alignment file as the other variants or are they on a separate file (bam/cram) ?

[dnil]

Right, always interesting with reference genomes. It looks like you are getting the IGV view and the correct locus, but the alignment is not showing. Is that correct?

[Jakob37]

Hi @Jakob37, where are your MT variants read from, same alignment file as the other variants or are they on a separate file (bam/cram) ?

They are stored in a separate bam file, here is how a sample can look in the case yaml (before adding owner):

---
owner:
family: 'hg002'
lims_id: '0000-00'
samples:
  - analysis_type: wgs
    sample_id: 'hg002'
    sample_name: 'hg002'
    mother: '0'
    father: '0'
    capture_kit:
    phenotype: affected
    sex: male
    bam_path: /access/wgs/bam/hg002_dedup.bam
    mt_bam: /access/wgs/bam/hg002_mito.bam
    d4_file: /access/wgs/cov/hg002_coverage.d4

Right, always interesting with reference genomes. It looks like you are getting the IGV view and the correct locus, but the alignment is not showing. Is that correct?

Yes, that is how I understand it!

[Jakob37]

OK, actually better I double check our MT BAMs before I say anything. I'll check ...

[northwestwitch]

OK, actually better I double check our MT BAMs before I say anything. I'll check ...

We touched that code quite recently so I think it might be a bug

[Jakob37]

OK! 🐛

Yes, looking at the MT BAM from the screenshots above, there are reads present.

[northwestwitch]

I can look into it!

[dnil]

It's likely not the chromosome naming or reference genome as such: we are using the same MT genome sequence with the good old 1kg hg19 (displayed on hg38) - more likely how we special case to make the tracks for it.

[northwestwitch]

@Jakob37 any specific error if you right click on the igv page and choose "inspect"?

[northwestwitch]

Other question: are you navigating to the MT variant from another variant (not on MT) or from the IGV browser created from the link on the case page?

[Jakob37]

I'll check!

[Jakob37]

Some warnings in the console, but these are also there for the non MT page:

Other question: are you navigating to the MT variant from another variant (not on MT) or from the IGV browser created from the link on the case page?

I am navigating: case -> Clinical SNV and INDELs -> (finding an MT variant in the list) -> IGV viewer button

[northwestwitch]

Ok, so the javascript shows no errors, just warnings. Thanks!

[northwestwitch]

I've tried to reproduce locally by setting the genome build of the demo case to 38 (your case has build 38, right? ) and modifying the VCF file of the SNV variants so that MT chrom is now M. The track shows .. Mmmm or MTTT :)

[Jakob37]

your case has build 38, right?

Yes, we only use build 38

and modifying the VCF file of the SNV variants so that MT chrom is now M. The track shows .. Mmmm or MTTT :)

;)

Strange! I'll check if I see the same if loading our hg002 case locally.

[dnil]

Also possibly worth trying out Scout 4.88.1 ie IGV.js 3.0.5 instead of 3.0.4. It shouldn't be that, but if you have like very deep mt bams and not subsampled I guess it might possibly be that region load termination thing anyway?

[dnil]

I've tried to reproduce locally by setting the genome build of the demo case to 38 (your case has build 38, right? ) and modifying the VCF file of the SNV variants so that MT chrom is now M. The track shows .. Mmmm or MTTT :)

@northwestwitch did you also try changing the aln to have chrM instead of MT? The demo aln looks like this: 😊

ST-E00269:73:HLNFFCCXX:2:1103:25499:65669       163     MT      1       60      13S138M =       229     376     AGACATCACGATGGATCACAGGTCTATCACCCTATTAACCACTCACGGGAGCTCTCCATGCATTTGGTATTTTCGTCTGGGGGGTATGCACGCGATAGCATTGCGAGACGCTGGAGCCGGAGCACCCTATGTCGCAGTATCTGTCTTTGAT      +????????I?I???II?I????I??II?I???IIII+???I?I?????????I?I??I???III???III?III?I???5????II???II?I?I5????I??I???II???'???????+?+????II??II????55????I?II??I MD:Z:138        PG:Z:MarkDuplicates          RG:Z:ADM1464A1.171015_HLNFFCCXX-11_XXXXXX.lane2 NM:i:0  AS:i:138        XS:i:86
ST-E00269:73:HLNFFCCXX:2:2113:11505:16041       99      MT      1       60      3S148M  =       142     292     ATGGATCACAGGTCTATCACCCTATTAACCACTCACGGGAGCTCTCCATGCATTTGGTATTTTCGTCTGGGGGGTATGCACGCGATAGCATTGCGAGACGCTGGAGCCGGAGCACCCTATGTCGCAGTATCTGTCTTTGATTCCTGCCTCA      ?????II?I????I?III?I???IIIIII??I?I?I???????I?I??I???III???IIIIII????????????I???I????II???II??????I???????????????I???II??I?????III???I?II??III???????? MD:Z:148        PG:Z:MarkDuplicates          RG:Z:ADM1464A1.171015_HLNFFCCXX-21_XXXXXX.lane2 NM:i:0  AS:i:148        XS:i:83

[Jakob37]

Also possibly worth trying out Scout 4.88.1 ie IGV.js 3.0.5 instead of 3.0.4. It shouldn't be that, but if you have like very deep mt bams and not subsampled I guess it might possibly be that region load termination thing anyway?

Hmm, yes, I am testing things out with v4.88.1. For some variants I get BAM file(s) missing. While for others I get the very same situation - i.e. the button is there but no reads.

Some debug printing of the values when the button is there of the data sent to the template:

scout_dev  | variant.chromosome: M
scout_dev  | case.mt_bams True

if you have like very deep mt bams and not subsampled I guess it might possibly be that region load termination thing anyway?

I haven't been in contact with that before. Is that something inside the JS-part?

(I'll continue after 🥗 )

[dnil]

Oh, so your chromosome is a plain M? Not chrM? Here is the entry from the hg38_alias.tab from IGV:

chrM	MT	J01415.2	NC_012920.1	chrMT

I guess it could be that then!

[Jakob37]

Oh, so your chromosome is a plain M?

Yes, that is correct! Not sure about the origin story, but now it is there by inertia.

Is the solution to rename our MTs? Or can IGV be configured somehow? (Guess something must have changed, as the issue seems to have appeared now)

[northwestwitch]

Is the solution to rename our MTs? Or can IGV be configured somehow? (Guess something must have changed, as the issue seems to have appeared now)

No no, no change of VCF files. If the problem is indeed the alias tab files, why does the demo with only M works locally? Are you also providing an index for the bam files? because the track is not added if you are missing the index, check the code here:

scout/scout/server/blueprints/alignviewers/controllers.py

Line 315 in 0bc179b

if None in [

[dnil]

Tentatively it is not the aln chromosome name, at least I could not reproduce it with the demo.

EDIT: just the GUI fooling me. I should have used the CLI also to change the case. It reproduces with M in the aln for one of the individuals.
'

[dnil]

No no, no change of VCF files. If the problem is indeed the alias tab files, why does the demo with only M works locally?

Did you really also try with M in the aln though? 😉 But you are right, I probably botched my attempt. It should not work. Let me recheck.

[dnil]

Oh, so your chromosome is a plain M?

Yes, that is correct! Not sure about the origin story, but now it is there by inertia.

Is the solution to rename our MTs? Or can IGV be configured somehow? (Guess something must have changed, as the issue seems to have appeared now)

Hm, consider asking IGV to include it in the alias file? That would be an instant fix. Some backstory and official reference claims might help, but it might be enough just that it seems very very unlikely someone would ever name some other human genome contig in their hg38 reference M.
Alternatively, copy their file, add an "M" and change the URL to somewhere you control, like a static served by scout. You can find the URL to change here:

scout/scout/constants/igv_tracks.py

Line 11 in 0bc179b

HG38ALIAS_URL = "https://s3.amazonaws.com/igv.org.genomes/hg38/hg38_alias.tab"

Or revert to IGV.js <3 I suppose, but you don't want that.

[dnil]

If we decide to approach IGV.js about it it might help to note that either MT or M in the bam just works anyway on desktop IGV. Which I guess has an additional fallback for the chr naming, much like IGV.js did before.

[northwestwitch]

Did you really also try with M in the aln though?

Right, I did not modify the alignment. It makes sense. We could try to open an issue on igv.js and see if they fix it, otherwise we can include the alias tab files in scout and serve them from the app (bit more cumbersome)

[northwestwitch]

Here --> igvteam/igv.js#1896

[Jakob37]

Nice job hunting this down 💪 Hope they are fine with updating it.