What Is The Meaning Of A Gene Ontology Annotation Having Two Tuples With Different Evidences?

Question

[davidechicco]

In the Gene Ontology databank, there are some annotations that have multilple entries, with different evidence.

For example, the annotation involving the Ryr2 gene and the GO feature GO:0006874 (cellular calcium ion homeostasis) show a tuple having evidence IEA, another having evidence ISS, and another having evidence ISO. This signifies that this annotation has a tuple Inferred from Electronic Annotation (IEA), another Inferred from Sequence or Structural Similarity (ISS), and another Inferred from Sequence Orthology (ISO).

In a general sense, what does this mean?

Is the existance of a valid annotation between Ryr2 gene and GO:0006874 feature trustworthy or not?

If yes, why the IEA tuple is still present?

If not, what is the correct meaning of annotations with different evidences?

Thanks a lot

Answer 1

[sarahhunter]

There are many different sources for GO term annotations for proteins and genes; some are manually added, others are automatically added. If you look at the GO annotations through a tool like Amigo (or QuickGO), you will see every term that has been applied, no matter what the evidence. To be honest, IEA, ISS and ISO are pretty much analagous to one another.

In databases like UniProt, they often have some sort of precedence filtering in place, so if a term is given by a manual annotation and by an automatic method, only the manual annotation will be shown. For your example, you can see that only 9 terms are associated and that most of them have some sort of manual association.

If something is IDA (Inferred from direct assay), you can bet your life it will have a reference associated and will be very trustworthy. However, the automatic methods can be equally accurate, it's just that the individual annotations to proteins haven't been vetted manually and therefore you might have slightly less confidence in them, unless there is other evidence to support them.

GO term evidences are explained here: http://www.geneontology.org/GO.evidence.shtml

There is a recent NAR paper about how the GO works and how different annotations are made: http://europepmc.org/articles/PMC3531070

How InterPro does GO mapping is explained here: http://europepmc.org/articles/PMC3270475