CODEWITHSUNDEEP
bioinformatics
Travelling Salesman
Travelling Salesman

Reputation: 2281

Gene Ontology term Level vs. Gene Ontology term Type

I am facing a problem in finding out the right definition of the level and the type of the Gene Ontology term.

I know that the Type of the GO term is related to isA graph without considering the partOf relationships. But I am still confused when I am having the following case:

When finding the level:

If I have GO:123 (assuming it's the root) pointing to GO:345 and to the term GO:567 with isA relations in both links. Also, GO:345 is pointing to the term GO:567 with partOf relationship.

Now, what's the level of GO:567? Is it 2 because the root is pointing to it? or is it 3 because the son of the root is pointing to it? How should I deal with such cases in both, level and type of the graph?

Upvotes: 1

Views: 1671

Answers (1)

Alex Stoddard
Alex Stoddard

Reputation: 8344

You may get more helpful answers taking this question to Biostar. Or perhaps the GO website and GO mailing lists.

The issues you raise arise directly out of the semantics of the GO. It is a directed acyclic graph (DAG) so nodes do not have a single definition of depth or level. You could choose to use min(depth) or max(depth) if you want one consistent definition but that may not be a good choice for your application, which you don't describe. Likewise the edges of the graph have different properties describing different biological relationships ("is a", "part of" and "regulates").

How you best processes these properties depends on what you want to accomplish or what queries you want to make of the data. Because the annotation density of GO terms to genes varies greatly across organisms and terms, you may be better off considering metrics that measure the informativeness of annotations in your particular context rather than 'depth' in the graph.

Upvotes: 1

Related Questions