search

midas-network / midas-data

An ontology for MIDAS data types
1 stars 4 forks source link

Influenza #52

Open LucieContamin opened 1 year ago

LucieContamin commented 1 year ago

Following the issue #49 ,

Influenza was added to the ontology with:

Would it be possible to update it to have:

It would allow us to have all type of Influenza (as we don't always know the detail) and to match Tycho ontology (https://www.tycho.pitt.edu/dataset/US.6142004/)

Please let me know if any issues or need more information, thanks!

harryhoch commented 1 year ago

We should discuss whether influenza, A, B, and other (variants H5N1, etc.) should be modeled separately.

hoganwr commented 1 year ago

Agree we should discuss. I added precisely what was requested.

On Thu, Jul 13, 2023 at 10:29 AM Harry Hochheiser @.***> wrote:

We should discuss whether influenza, A, B, and other (variants H5N1, etc.) should be modeled separately.

— Reply to this email directly, view it on GitHub https://github.com/midas-network/midas-data/issues/52#issuecomment-1634452338, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJR55TMWXRCYILSEAAYMTDXQAH65ANCNFSM6AAAAAA2CCSI2I . You are receiving this because you are subscribed to this thread.Message ID: @.***>

hoganwr commented 1 year ago

With respect to influenza, anyway. I think I varied from request on HIV because of peculiarities of the NCBI Taxon resource.

On Thu, Jul 13, 2023 at 11:02 AM Bill Hogan @.***> wrote:

Agree we should discuss. I added precisely what was requested.

On Thu, Jul 13, 2023 at 10:29 AM Harry Hochheiser < @.***> wrote:

We should discuss whether influenza, A, B, and other (variants H5N1, etc.) should be modeled separately.

— Reply to this email directly, view it on GitHub https://github.com/midas-network/midas-data/issues/52#issuecomment-1634452338, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJR55TMWXRCYILSEAAYMTDXQAH65ANCNFSM6AAAAAA2CCSI2I . You are receiving this because you are subscribed to this thread.Message ID: @.***>

LucieContamin commented 1 year ago

Looking at the data, I think we should at least have these possibilities:

We can maybe add a specific variant (for example H5N1) if necessary. Thanks,

hoganwr commented 1 year ago

NCBI Taxon ID 11309 is not a parent class of both Influenza A and Influenza B. In fact, the associated label is "unidentified influenzavirus". Inappropriate mixing of epistemology and ontology aside, it cannot serve the function we wish it to.

Here is the classification from NCBI:

Orthomyxoviridae Alphainfluenzavirus Alphainfluenzavirus influenza Influenza A virus Betainfluenzavirus Betainfluenzavirus influenza Influenza B virus

hoganwr commented 1 year ago

And of course GitHub stripped out all my indentations!!!

hoganwr commented 1 year ago

Orthomyxoviridae

hoganwr commented 1 year ago
Screenshot 2023-08-02 at 4 01 29 PM

Trying a screen print from the NCBI Taxonomy Browser, but I am not confident...

LucieContamin commented 1 year ago

Thanks for information,

To provide more information: We wanted to be able to represent

First, the currentDisease - Infectious disease - Influenza (http://purl.obolibrary.org/obo/MONDO_0005812) has exact match with Influenza with non-respiratory manifestation (https://snomedbrowser.com/Codes/Details/61700007) which I don't think will work in our situation, so I propose we change to Influenza SNOMED 6142004 (see below).

For the pathogen associated with the disease "Influenza", as the "Influenza" can represent case from either A, B or unknown or both, it is difficult to assign a pathogen. Tycho has an unidentified influenza virus used for "Influenza" disease. However, following @hoganwr comments, the taxonomy of the pathogen seems to cause issue. Here I copied summary of the taxonomy (to have all information in one place):

NCBI Taxon ID 11309 is not a parent class of both Influenza A and Influenza B. In fact, the associated label is "unidentified influenzavirus". Inappropriate mixing of epistemology and ontology aside, it cannot serve the function we wish it to

I understand the remark. I don't think something other than "unclassified" pathogen can be use here as we don't know how to "classify" the associated pathogen with the data associated with Influenza, as it can be (A and B) or one strain only but not tested. However, as FLU A and B are the main viruses that routinely spread and cause the seasonal epidemics each year (CDC), I think for some data, we can assume we have both (A and B) in our "Influenza" datasets.

So, I have 2 propositions:

Proposition I (use the imperfect unidentified pathogen):

Proposition II (have no pathogen specific to Influenza and have A and B as children):

Thanks!

harryhoch commented 1 year ago

@hoganwr -- discussed with @LucieContamin.

We agreed that proposition II is preferred, diseases and pathogens should be in distinct, mirrored subtrees (with appropriate relationsships between them).

LucieContamin commented 1 year ago

Just to clarify my proposition II, I was thinking to have in the ontology:

and:

hoganwr commented 1 year ago

I will add the virus classes needed. The "Influenza due to Influenza A" and "Influenza due to Influenza B" part is going to require some extra thinking/work. No OBO ontology has those classes, and I'm not set up at the moment to pull in IRIs from non-OBO artifacts.