From ReactomeWiki

How to indicate that an entity or event is part of a disease process

If a protein variant (or other entity) or event is associated with a disease, this should be indicated by filling in the Disease attribute in the event or Entity.

For example, to specify that the protein EGFRvIII is associated with adult glioblastoma multiforme

1. Browse the EBI disease ontology at here.
   • The DOID for adult glioblastoma multiforme, the cancer expressing EGFRvIII, is 3075.
2. Right click on disease field in the EGFRvIII EWAS record and select "Add".
3. Browse gk_central to determine if the DOID is there.
   • If it is, select it.
   • If it is not, click on New Instance.
4. Set the Identifier to 3075 and press Enter.
   • You will be asked if you want the tool to fetch the information from the EBI ontology for the specified disease instance.
5. Select “Yes”.
   • Values for the specified identifier will populate automatically.
6. Press OK. When the Disease window closes, press OK again.

The Disease slot for an entity is multivalued, so if you are curating a disease process that involves a variant protein for which the appropriate EWAS already exists (i.e., the same referenceGeneProduct, same compartment, exactly the same modifications), add disease identifier of interest in the disease slot on the existing EWAS and re-use it in your annotation – do not create a new EWAS identical to the old one except for its “disease” attribute.

Note - A disease attribute should also be assigned to events involving a disease variant EWAS.

Caveats in assigning disease to sets/events

All EGFR kinase domain mutants behave in a similar way when it comes to downstream signaling, and are therefore grouped in a set. However, while the majority of set members can be found in lung cancer, one particular member has been found exclusively in glioblastoma, and some, but not all, of the lung cancer mutants have been found in breast cancer, thyroid cancer, gastric cancer, etc. If I assign lung cancer disease value to sets and events, it’s a generalization that may be justified by the fact that kinase domain mutations of EGFR and related events really are the hallmark of lung cancer. However, applying other disease values to sets and events involving EGFR KD mutants, e.g. breast cancer, thyroid, glioblastoma, where this generalization is true for only a small subset of EGFR KD mutants, doesn’t feel right.

Q:My question is whether to restrict disease attribute to EWAS only?

If not, what should be the rule for assigning disease attribute to sets and reactions

A:The disease attribute is designed as an optional attribute. So in your EntitySet case, you can just leave it empty for the EntitySet instance, but assign it to each individual EWAS instances that you know. Also it is true for Events. If you think an EntitySet or Reaction can occur in cancer cells or tissues only, you may add the generic "Cancer" disease term. Otherwise, please leave it empty there. A software program should be able to find these EntitySets and Reactions if any member of any pathway participant has been annotated for a specific disease already.