search

OmniSearch / ncro

Non-Coding RNA Ontology
Creative Commons Attribution 4.0 International
5 stars 3 forks source link

miRNA_gene_family #9

Open alanruttenberg opened 8 years ago

alanruttenberg commented 8 years ago

All the miRNA_gene_family are information artifacts, but they should probably be classes of (material) miRNA, and superclass of their respective (RNA) members. If they are, then the "gene family" is a misnomer - they are really RNA families. If they are really gene families, then they should still be material entities, and the relation of them to the RNA isn't aboutness - it is creation by transcription.

Huang-OMIT commented 8 years ago

Under "miRNA" we have "human_miRNA" that includes all human miRNAs as its subclasses. On the other hand, each and every human miRNA has a relation with "miRNA_gene_family." For example: hsa-mir-100 is_classified_into_gene_family_group miRNA mir-10 mir-10 is_about_grouped_miRNA hsa-mir-100

Such a definition and 2-way relation seems clear to me.

alanruttenberg commented 8 years ago

The instances of human_miRNA are little individual miRNA molecules. What are the instances of miRNA_gene_family?

Huang-OMIT commented 8 years ago

The main purpose of miRNA_gene_family is to classify different miRNA molecules. We do not have to have instances for miRNA_gene_family. For example, instances may be some families in an individual person?

alanruttenberg commented 8 years ago

BFO doesn't recognize classes without instances. Every assertion is an assertion (possibly quantified) about instances.

Having the gene families be superclasses of mirNA classes means that the family's instances would be molecules too, which makes sense and is the same way PRO organizes its hierarchy. We wouldn't name them with "family" as the instances are not families, but proteins or aggregates of proteins.

In a BFO ontology, we would represent family_role, realized as as something like living as a family (fill in the details). The bearer, i.e. family would be an aggregate of humans who participate in the realizing process. So the relation of a people to a human family is similar to the relation of a phone to its parts.

Alternatively you could have information entities instances (say 'mir-125-classifier') be members of a class of miRNA_gene_family_classifiers, and then each one of those instances would be about all the members in the union of hsa-mir125, mmu-mir-125, etc.

Protein ontology chooses the classes approach as they define the the 'gene level' protein families as those that are evolutionarily related to each other. It has been the same in a number of other ontologies. For instance there are two views of taxa. In one view we have homo sapiens being an individual with parts that the people. In the other we have homo sapiens be a class with people as instances. Our rendering of NBCI taxonomy understands it as the latter.

Huang-OMIT commented 8 years ago

Did you mean to use the "family_role" solution? If so, agreed. Please make corresponding changes.

linikujp commented 8 years ago

I think my original idea of miRNA_gene_family is an information artifact. The 'family role' doesn't reflect the biological reality.

A gene family/miRNA family is more a grouping concept than a role.

alanruttenberg commented 8 years ago

I don't think the human family example is right for this case - in the case of a family the actors can work together. In the case of the miRNA the miRNA is spread across the world and does not work in concert.

I lean towards the PRO approach. mir-125 is the class of miRNA transcribed from a gene with a common evolutionary origin. See http://pir.georgetown.edu/pro/ - PRO calls this the "Family level distinction"

So , e.g, mir-125 would be a class, subclass of 'genetic non-coding RNA', and hsa-mir-125 and mmu-mir-125(the 'gene level distinction') would be classes, subclass of mir-125.

Huang-OMIT commented 8 years ago

Like this solution with one minor change: hsa-mir-125 will be subclasses of BOTH mir-125 AND human_miRNA.

Anyway, this issue will not affect our paper and can be further discussed.

alanruttenberg commented 8 years ago

That's not a problem. We wouldn't make them both parents in protege. We would define human miRNA as miRNA and only in taxon human. The hsa-mir-xxx would also have an axiom only in taxon human. The reasoner would then infer the superclass relationship.

linikujp commented 8 years ago

Where will you put the 'gene level distinction' in the ontology?

alanruttenberg commented 8 years ago

gene-level distinction isn't an entity in the ontology. It is a metaclass. I think PRO has an annotation property.