Gene To Test Machine Learning Algorithm

Question

[9606]

I'm developing an application to predict new functional annotations on Gene Ontology.

So far, i used ROC curves and likelihood to evaluate the model performance.

Now I would like to do some more advanced test, in detail I want to choose a gene, take its set of annotations in an old version of Gene Ontology, give them as input to the algorithm and compare the predicted annotations with the ones present in a recent version of Gene Ontology.

Unfortunately my biologica background is very limited, so i would ask you if you can suggest my some genes/protein (if it is possible from HomoSapiens or MusMusculus) that is "well known", that has a consistent and rich set of annotations in GO.

Thank you.

Answer 1

[Zev.Kronenberg]

Refseq genes are usually considered high quality. Also if the go term is Inferred from Electronic Annotation ( IEA ) I would avoid using it.

Sounds like a cool project.

link to refseq:

http://www.ncbi.nlm.nih.gov/RefSeq/