Support bedMethyl format.

[ArtRand]

Hello,

Jumping off from: https://github.com/nanoporetech/modkit/issues/43

I develop a tool for secondary analysis of base modification data, modkit. One use case it fulfills is to generate "bedMethyl" tables. The schema of these tables have been described by Encode, modbam2Bed, and again my modkit. In principal, they're all BED files, however to load the tables generated by modkit, I need to use spaces to delimit the fields after the 10th field. Perhaps not surprisingly, this causes a bunch of downstream headaches, but it was important to me that the output of the tool can be immediately loaded into IGV.

This issue is to track the ability of IGV to load bedMethyl tables such as those produced by modkit when only tabs are used to separate the fields. (Modkit has an --only-tabs flag to allow this).

Please find a "tab-only" version output here: https://raw.githubusercontent.com/nanoporetech/modkit/master/tests/resources/modbam.modpileup_nofilt.bed

The reference for this BED can be found at: https://raw.githubusercontent.com/nanoporetech/modkit/master/tests/resources/CGI_ladder_3.6kb_ref.fa

[jrobinso]

OK this format will be supported in release 2.17.0, targeted for the end of October. To test or use it before then you can download our snapshot build from here: https://software.broadinstitute.org/software/igv/download_snapshot

If you have a "bigBed" version please point me to it, or upload it here. It should just work but it would be good to test it.

[ArtRand]

@jrobinso Sounds good. I'll add bigBed output to modkit, it's a feature that's been requested before.

[jrobinso]

@ArtRand I have a question on your sample file, it seems to have more columns than the spec defined, for example

chr20   22613919    22613920    h   2   +   22613919    22613920    255,0,0 2   0.00    0   2   0   0   0   0   0

The spec defines 2 extra columns

• Coverage, or number of reads
• Percentage of reads that show methylation at this position in the genome

The extra columns should be harmless, just noting this.

[jrobinso]

OK this is set for the next release, 2.17.0, sometime in October.

[ArtRand]

@jrobinso great I'll give it a shot!

Btw. I count 18 columns..?

chr20   22613919    22613920    h   2   +   22613919    22613920    255,0,0 2   0.00    0   2   0   0   0   0   0
1: chrom - chr20
2: start - 22613919
3: end - 22613920
4: mod-code - h
5: valid cov - 2
6: strand - "+"
7: start, again  - 22613919
8: end, again -  22613920
9: color - could take better advantage here - 255,0,0
10: coverage, again - 2
11: %-mod - 0.00
12: n_modified - 0
13: n_canonical - 2
14: n_other_mod - 0
15: n_delete - 0
16: n_fail - 0
17: n_diff - 0
18: n_nocall - 0

[jrobinso]

OK, but only the first 11 are specified here: https://www.encodeproject.org/data-standards/wgbs/. If you want to include the rest in the popup text we can do that, but there is nothing in the bed file itself to tell us what they are. With bigBed you can define this in autosql, and in fact every file could have a different set of extra columns, with "bed" type files you just have to know.

Until release the new support is only available in the snapshot build for IGV desktop, and https://igv.org/app-test for the webapp.

On Tue, Sep 26, 2023 at 5:52 PM Arthur Rand @.***> wrote:

@jrobinso https://github.com/jrobinso great I'll give it a shot!

Btw. I count 18 columns..?

chr20 22613919 22613920 h 2 + 22613919 22613920 255,0,0 2 0.00 0 2 0 0 0 0 0 1: chrom - chr20 2: start - 22613919 3: end - 22613920 4: mod-code - h 5: valid cov - 2 6: strand - "+" 7: start, again - 22613919 8: end, again - 22613920 9: color - could take better advantage here - 255,0,0 10: coverage, again - 2 11: %-mod - 0.00 12: n_modified - 0 13: n_canonical - 2 14: n_other_mod - 0 15: n_delete - 0 16: n_fail - 0 17: n_diff - 0 18: n_nocall - 0

— Reply to this email directly, view it on GitHub https://github.com/igvteam/igv/issues/1396#issuecomment-1736495926, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAHD2HFKAKACTEE3J3CXBNDX4N2FPANCNFSM6AAAAAA5GXSPHM . You are receiving this because you were mentioned.Message ID: @.***>

[ArtRand]

I thought you meant the modkit spec. As you can see we've mostly taken the encode spec as "inspiration". I'll move to output bigBed from modkit.

[jrobinso]

Ahh, o.k., in one of these tickets someone directed me to the Encode site. I'll update to use this as the spec: https://github.com/nanoporetech/modkit/blob/master/book/src/intro_bedmethyl.md

[jrobinso]

I added support for all columns in the javascript version, will do desktop soon, but first have a look at the attribute names. These only appear in popup text, and can be whatever you like, just post corrections here.

[ArtRand]

@jrobinso This is great, thanks. I'm assuming I can modifiy modkit so that the output always "all-tabs" (previously all-tabs required a flag)? I can test on both platforms, of course. BigBed output may not come in the next release - but a fast follow since it will be nice for things like DMR output as well.

[jrobinso]

For IGV purposes BigBed is not so important now that bedmethyl is a recognized format. Its nice for unknown bed formats because the # of defined fields and names of the extra columns can be discovered in the BigBed autoSql. Also good if your files get large.

[KloostermanJoukje]

Hello,

I am wondering if the release is ready? I am having the same issue as @ArtRand as described by his first message. I have a Bedmethyl file i would like to load into IGV.

[jrobinso]

@JoukjeKloosterman Working on it, its a big release, however you can get a preview here: https://igv.org/doc/desktop/#DownloadSnapshot/