Annotation Conf. Call, January 22, 2013

From GO Wiki
Jump to: navigation, search

Agenda

Introduction of the PomBase Matrix Project (Val)

https://sourceforge.net/apps/trac/pombase/wiki/MatrixProject

Slides: http://wiki.geneontology.org/index.php/File:Matrix_talk.pdf

For groups moving to Protein2GO - consideration of IEP annotations

We want to make a point in the documentation about how IEP annotations will be represented when annotating to UniProt accessions. Does the following sound reasonable to groups?

"UniProt rarely annotates to expression pattern (IEP), but when we do, we ensure that the experiment shows protein expression data, we do not annotate gene expression data. However, UniProt currently does display IEP annotations from other groups mapped to UniProt accessions.

Groups who annotate gene expression data using IEP should understand that they may be making a different statement to what was intended when associating the annotation with a UniProt accession. They may wish to exclude their IEP annotations when supplying their annotations to UniProt for the initial integration into Protein2GO and then append them to the association file UniProt supplies back to them before submitting their annotations to the GO Consortium."

Discussion

  • IEP: Should we make granular evidence code for protein and gene expression?

Donghui's question- We use IMP evidence to indicate gene or protein mutant and we don't make the distinction. Along those lines, using IEP to mean protein or gene expression should be fine.

The consensus was that we should make more granular IEP evidence codes for protein and gene expression in ECO for future use in annotation, but we should not worry about users making inferences about existing IEP annotations. However, we should improve the documentation to state that different annotating groups use different identifiers and no inferences should be made as to whether the annotation concerns a gene, RNA or protein.

  • Judy: multiple UniProt IDs for various forms of the same protein (human cdk2 example) which results in GOannotations for all these IDs and it skews enrichment and other analysis. Rachael suggested that we bring this up with Claire.