10 June 2014 PAINT Conference Call
Minutes from PAINT call - June 10, 2014
- Pascale reported to Suzi that when you 'open' a family, there are two small bugs with evidence .txt file PAINT:
1. Everytime we save, it adds a new #notes line. It's OK, it just looks funny. Not sure if it's a difficult thing to fix. See here for an example: http://pantree.org/tree/familyCuratorNotes.jsp?accession=PTHR10516
2. The WARNINGs in the red box are missing returns at the end of lines, which makes it really hard to read.
3. There is a bigger bug on Windows: when we try opening a saved family, it doesn't open (pascale will send the error message in the PAINT console). Works fine on Macs.
1. Annotating a family with evidence from a single species: Karen asks whether she should annotate a family in which there is only yeast evidence, but the family is well conserved and the process is believed to occur in other species in which the protein is found [Karen: can you specify which family you were talking about ? ]
-> yes, we agreed that this is where PAINT is very valuable
2. Annotating 'drug binding' function
- From Pascale: I am annotating (or rather, reviewing) a family of FK506 binding proteins, PTHR10516. SInce FK506 binding is mentioned A LOT everywhere in the literature, I propagated this annotation.
- Val disagrees with this annotation: "PomBase wouldn't make this annotation because it isn't describing a function or process which occurs in wild type yeast. Its a 'condition' which produces signalling problems. I don't think its good practice to include this binding activity in GO annotation describing normal physiology. We might say that a specific enzyme could be inhibited by his compound, but anything else would be described using phenotype annotation."
- See discussion here: https://sourceforge.net/p/geneontology/annotation-issues/852/
- We agreed that if the term was legitimate, and the annotations were legitimate as well, then this should be propagated, since it's been widely tested and there are many paper about FK506 binding across multiple species (including cerevisiae: http://www.yeastgenome.org/locus/S000002927/go), that seemed an important annotation at the primary annotation level, and was therefore propagated.
- Question 1: Is this a legitimate GO term ?
- Question 2: Should binding to *any* compound be annotated ? Or does it need to be 'physiologically relevant'? And how does one assess physiological relevance ? If a protein is a target of a drug, can you annotate that it binds to the drug ? Or are annotations limited to 'defense' proteins ?
- Note also: In PAINT we don't usually capture bindings. This is a special case because there is a lot of literature describing the binding; thus it seems of interest to the scientific community.
- Action: Rama to bring this as a discussion for the Ontology and annotation groups.