mondo-base.obo: MONDO IDs do not match between $xref and $id

[bschilder]

I'm using the "mondo-base.obo" object read into R as an ontologyIndex object.

In working with this object, I've noticed some unexpected features. Namely, the MONDO IDs in the $xref slot do not seem to match up with those listed in the$id slot. I'd expect that all MONDO IDS in any of the slots should at least be present in the $id slot.

Mondo term (ID Label)

xref_ids.csv

Bug/Typo/Error description

remotes::install_github("neurogenomics/HPOExplorer")
mondo <- HPOExplorer::get_mondo()

# Only 18 MONDO IDs are missing names, which is ok
>     sum(is.na(mondo$name))
[1] 18
# 10273 MONDO IDs are missing definitions; less ok but still manageable
>     sum(is.na(mondo$def))
[1] 10273
> 

# Get MONDO IDs listed in `$xref`
> xref_ids <- unique(names(unlist(mondo$xref)))
# Get total number of xref IDs
> length(xref_ids)
[1] 101231
# Almost all of these xref MONDO IDs (~100k) are missing in the `$id` slot! in theory all IDs should be in the `$id` slot
>     sum(!xref_ids %in% names(mondo$id))
[1] 98468
# Same missing rate in the name slot
>     sum(!xref_ids %in% names(mondo$name))
[1] 98468
# Same missing rate in the def slot
>     sum(!xref_ids %in% names(mondo$def))
[1] 98468
> 

Your nano-attribution (ORCID)
https://orcid.org/0000-0001-5949-2191

Thanks in advance for your help.

[bschilder]

Hey, I think this may have gotten buried amongst the other Issues, but wanted to check in and see if anyone has looked into this yet. @twhetzel

Thanks for your help!

-Brian

[twhetzel]

Hi Brian - thanks for the ping on this issue. I've been looking into this and have a few questions and comments.

What is the source of the Mondo ontology file used? I've looked at the number of terms without labels or definitions in the latest release of Mondo and do not get the same count as mentioned in the initial post.

In the "xref_ids.csv" file, the Mondo IDs vary in length from 13, 14, and 15 characters long while they should be 13 characters long (MONDO: and then a 7 digit number). Amongst the IDs that are 13 characters long, there are at least a few obsolete terms in the file, e.g. MONDO:8000034. Deprecated terms in MONDO have the annotation owl:deprecated with value true [1] and of the Mondo terms with xrefs in the latest release, 3,210 of these are obsoleted.

For other valid IDs in the "xref_ids.csv" file, e.g. MONDO:0000005, what are the values you get for $xref and $id?

[1] https://mondo.readthedocs.io/en/latest/editors-guide/merging-and-obsoleting/#obsolete-a-class-manually

[twhetzel]

@matentzn have you taken over this ticket since you created a meeting with Brian?

[matentzn]

I Never saw this, and I didn't realise @bschilder was interested in Mondo at all! Yep I can discuss it with him when I meet him.

[twhetzel]

Oh, I thought your ping to him here (neurogenomics/RareDiseasePrioritisation#33 (comment)) was related to Mondo as well

[matentzn]

I guess now it it but originally no, it was only referring to uPheno!

[matentzn]

We determined this is not a Mondo related problem, but related to the r toolkit!

[bschilder]

Indeed, it seems something strange is going on within ontologyIndex::get_OBO which HPOExplorer uses to import the latest MONDO ontology from GitHub releases into R. Specifically, this OBO file:
https://github.com/monarch-initiative/mondo/releases/download/v2023-09-12/mondo-base.obo

For example, the Mondo IDs within $xref seemed to be a mix of obsolete and completely made up MONDO IDs! (according to @matentzn who scanned through the CSV of missing MONDO IDs i attached above).

While I'm still trying to sort out the exact reason for this issue, I think ontologyIndex::get_OBO is a pretty good lead. I'll reach out to those authors and let you know what the outcome is.

[twhetzel]

Thanks both for the update. Please see other issues with the IDs as mentioned earlier #6873 (comment)

[matentzn]

Yeah I confirmed that with Brian in the meeting, many ids don't even exist. He will look into it!

[bschilder]

For other valid IDs in the "xref_ids.csv" file, e.g. MONDO:0000005, what are the values you get for $xref and $id?

@twhetzel here's a reprex you can quickly run in R:

> if(!require("ontologyIndex")) install.packages("ontologyIndex")
> mondo <- ontologyIndex::get_OBO("https://github.com/monarch-initiative/mondo/releases/download/v2023-09-12/mondo-base.obo", extract_tags = "everything")

 > mondo$xref["MONDO:0000005"]
$`MONDO:0000005`
[1] "OMIMPS:203655"

> mondo$id["MONDO:0000005"]
  MONDO:0000005 
"MONDO:0000005" 

[bschilder]

Tagging who i think may be the author of ontologyIndex @daniel-jg

• https://github.com/cran/ontologyIndex/blob/master/DESCRIPTION

[twhetzel]

@bschilder that looks correct for MONDO:0000005. Since Nico mentioned that it was determined the issue is with the R toolkit, can this ticket remain closed since it is not an issue with Mondo?

[bschilder]

@bschilder that looks correct for MONDO:0000005. Since Nico mentioned that it was determined the issue is with the R toolkit, can this ticket remain closed since it is not an issue with Mondo?

Sure thing, I've already started working on alternative methods for converting IDs across ontologies using some of Monarch's resources. I'll keep your team posted on how things progress:
https://github.com/neurogenomics/KGExplorer/blob/29eccbbd33fd18d9ce85b0ae72b47d485d97faee/R/map_mondo.R