Conf. call with IDO 3-23-12

Friday, March 23, 2012, 2PM ET

Agenda/minutes for conference call with PO and IDO curators.

In attendance: Albert Goldfain, Lindsay Cowell, Pankaj Jaiswal, Justin Elser, Ramona Walls

Back to Plant Disease Ontology main page.

Questions arising from reviews

Real data or problems in modeling disease

Is there an example/data set we can use as a test case?

e.g., Gene Associations for Trait Term "rice bacterial blight disease resistance" from species "Oryza sativa" in Gramene

Perhaps mine GO annotations for a specific pathogen.

We can use the data in Gramene as a test case, and PJ could curate some papers for pathogen expression if we need them. We don't need to worry about this too much for the ICBO paper, but it will be important going forward.

Albert: IDO-PD is a reference ontology, not an application ontology, so having a specific use in mind is less important. Application ontologies can be built off of the IDO-PD.

access to ontology

RW set up SVN folder for draft on PO SVN: http://palea.cgrb.oregonstate.edu/viewsvn/Poc/trunk/ontology/collaborators_ontology/plant_disease/

Should we put a link to this in the paper?

Need to set up link for SVN externals.

JE also set up an AmiGO browser with a test version of the stress ontology: http://stressontology.cgrb.oregonstate.edu/cgi-bin/go

Potentially problematic terms from IDO

"The authors point out that the use of ‘clinical’ in OGMS / IDO and the definition of ‘symptom’ are problematic when building a plant disease ontology. This is (to me) one of the most interesting aspects of the paper. Have the authors tried to come up with more general definitions that would be universally applicable? Or is this not possible (why not)? It seems that the present structure works well for the plant ontology regardless, so I am not asking to revise the decision to build on OGMS / IDO. Rather I would be interested in a more detailed treatment of these problematic areas."

Examples of terms:

IDO:transition to clinical abnormality : A process by which a part of an organism or something contained in an organism becomes clinically abnormal.

IDO:subclinical infection: An infection that is part of an asymptomatic host.

What does clinically abnormal mean? Can we apply this same concept to a plant?

Could we call this asympomatic infection, with "subclinical infection" as a synonym?

We agreed that the meaning of these two terms is compatible with plants, the only problem is the use of the word clinical. We can say something about that in the paper.

Description of clinically abnormal from http://ontology.buffalo.edu/medo/Disease_and_Diagnosis.pdf, p2 bottom second- top third columns:

When we say that some bodily feature of an organism is clinically abnormal, this signifies that it:

i) is not part of the life plan for an organism of the relevant type (unlike aging, pregnancy or menopause),

ii) is causally linked to an elevated risk either of pain or other feelings of illness, or of death or dysfunction, and

iii) is such that the elevated risk exceeds a certain threshold level [5].

This treatment of ‘abnormal’ is distinct from those statistical treatments which do not take account of the overlap in the distribution of test results between

normal and abnormal populations or of normal distribution extremes. What are standardly called ‘normal variants’ (for example a left lung with three lobes) do not satisfy criteria ii) and iii).

OGMS:symptom: A quality of a patient that is observed by the patient and is hypothesized by the patient to be a realization of a disease. (note that "patient" is not in the IDO)

Def. of symptom from Roberts and Boothroyd "Fundamentals of Plant Pathology": A visible expression by a suscept of a pathologic condition.

OGMS:disease: A disposition (i) to undergo pathological processes that (ii) exists in an organism because of one or more disorders in that organism.

Def. of disease from Roberts and Boothroyd: Harmful physiological processes caused by continuous irritation of a plant by a primary causal agent. It results in morbific cellular activity and is expressed in characteristic pathological responses called symptoms; continuous malfunctioning.

Def. of disease from Manners "Principals of Plant Pathology": harmful deviation from normal functioning of physiological processes. This definition allow malfunctions caused by non-living agencies such as mineral deficiency to be classed as diseases: such diseases are sometimes known as disorders...

There is some difference between how a disease is viewed in OGMS versus in plants. May be viewed more as a process in plants, while in OGMS it is a disposition. In OGMS, you can say that someone has a disposition (to a particular a disease) if they are infected but not yet showing symptoms. A plant may also be infected and show no symptoms, but you would not say it has a disease until it has some "symptom" (physiological or morphological).

Difference between "disease" and "disease course".

There was also some discussion of the difference between a symptom (reported by patient) and a sign (reported by a third party). You could say that plant only has signs.

Perhaps the best solution to this problem is not to classify symptoms (or signs) as is_a symptom, but to classify them more essentially by what they are and then add relations like "symptom_of" and "has_symptom". These relations are being developed by OGMS.

For example:

• a leaf lesion is_a portion of necrotic tissue that is part_of a leaf

• rice bacterial blight disease has_symptom leaf lesion

Important to remember that "symptom_of" and "has_symptom" are not invertible, so you need to only use the relation that is always true (will probably most often be "has_symptom" because most symptoms are not unique).

Relations

"Several of the IDO-PD relations sound like they should be in IDO proper. Will they be communicated back? It also seems that these relations are “shortcut relations” for more complex constructs in BFO/RO. Have the authors explored making this connection explicit?"

From manuscript:

has_causal_agent: This relation is used to indicate the causal agent of a disease or other stress. It has two subtypes: has_causal_biotic_agent and has_causal _abiotic_agent. In the IDO-PD, the domain is a plant infectious disease and the range is an organism that is bearer_of an infectious agent role.

Developing for OGMS - a way to link dispostions with material entities that produce those dispositions to link a disease to its underlying disorder. has_material_basis - does not imply a causal relation but the material basis for the disease.

Definition (under development): 

not invertible

d has_material_basis m iff for all times t

d instanceOf bfo:disposition AND

m instanceOf bfo:material entity AND

i instanceOf bfo:independed continuant AND

         i has_disposition d at t implies i has_part m at t

IDO is working on a relation for "has_infectious_agent".

For example, could say that malaria has_infectious_agent Plasmodium, and also that malaria has_material_basis Plasmodium.

The reason has_causal_basis was developed, is because you cannot use "has_infectious_agent" for non-infectious diseases (e.g., abiotic stressors) in plants. However, we may still be able to use has_material_basis in this case.

primary/alternative_host_of: Two relations used to indicate whether a plant species is the primary or second-ary/alternative host of a disease, e.g., X. oryzae infected O. sativa primary_host_of X. oryzae infectious agent. The domain is an organism that is a subtype of Viridiplantae and is the bearer of a host role, and the range is an organism that is the bearer of an infectious agent role. Secondary hosts may be wild plant species that provide a place for the disease to survive when the crop host is not present or during a different part of the disease organism’s life cycle.

has_symptom: This relationship type is used to indicate the associated symptoms/traits that are evaluated to ascertain the manifestation of a stress or disease, e.g., rice bacterial leaf blight disease has_symptom leaf lesion.

found_in: This relationship is used to indicate the geograph-ical location of a host and/or pathogen species, e.g., X. oryzae found_in X. oryzae range. Instances of a disease or organisms also may have a found_in relation to an instance of a geographic region.

Trait Ontology

"While most ontologies mentioned are introduced a bit, the ‘trait ontology’ is only explained as “describing traits” (manuscript text) and dealing with “plant traits” (Table 1), which is not helpful to someone like me who is unsure what the scope of that ontology is."

Can PJ provide a one sentence description of the scope of the TO?

A trait is something you observe - a sign or symptom. Like a phenotype. Leaf lesion is a trait.

The thing that you observe, not your observation.

PAMGO

Other issues

Scope

Do we want to cover all green plants or just land plants?

IRI's

RW set up Protege so that IRIs follow the OBO Foundry (http://purl.obolibrary.org/obo/idopd.owl/IDOPD_0000000) but we have no special permission from OBO to do this, and right now these links are meaningless.

Is there a temporary alternative to use until the ontology is ready for submission to OBO Foundry?

Syncing with other ontologies

IDO

Multiple comments in IDO could cause problems with working with OE.

Can we create a subset of IDO with only terms that are relevant to plants?

TO

The main issue will be from TO -- using IDOPD terms to logically define TO terms.

Mireoting

Could Lindsay or Albert provide some help to RW on how to import terms using Mireot or other techniques.

How did you import terms from NCBI? Just re-use them/enter by hand?

Importing organisms

NCBItaxon ontology versus UBio id

What exactly do each of those stand for?

"A taxonomic classification of living organisms and associated artifacts for their controlled description within the context of databases."

From UBio: "NameBank is a biological name server. It stores and serves NameBankIDs via Life Science Identifiers, a persistent and distinct identifier. Each NameBank LSID returns a NameBank object which contains information about the NameBank record."

Neither of these resources say anywhere that their ids represent organisms of a particular class.

Wiki page

Should this be part of the PO wiki, should we create a separate wiki for IDO-PO?

Versioning and release policies

(Probably don't need to worry too much about these yet)

• OWL versus OBO
• Pre-reasoned version

• How to include imported ontologies with downloads.