Column labels in results and dowload can be confusing

[ValWood]

When you filter on "gene products" the column heading with the "gene name" is called "Acc"

I searched on "biological_process"
selected the gene products link,
filtered on H. sapiens to get all gene products annotated to BP node
Filtered on type "protein"

The confusing thing is that there is not a "gene/name product" column heading in this view...the column is called "Acc" but sometimes it is containing the "gene product" label. Sometimes a UniProt type Identifier
e.g.
KDM4A_HUMAN

filter on "SAA1" and you will see what I mean.
You see
SAA1
and
SAA1_HUMAN

This means that the "get products" link filtering does not give you a unique list of gene products because SAA1 is represented twice in this list.

Mainly now (as I think Chris is onto the redundency issues) I can't figure are why the column is called "Acc" here and not "gene/product" (as it is called in the link to "annotations"
It seems that the "gene/product" column is the name that biologist will usually expect to see.

I was also pretty confused by the labels for the download options. I think they need explaining more clearly (maybe how they match the GAF, or exactly what they will contain?)
Acc(id)
Acc(bioentitiy)
Label(bioentity_label)
name(bioentity_name)

Is there any document describing these?
I tried all of these and none of them seems to consistently supply me with the label which I was looking for DB Object Symbol (column 3 in the GAF)?

[kltm]

There seem to be three issues in this one ticket; it's best to split them up as fixes would be independent:

• The column names aren't always great and can be confusing.
• There are duplicate GPs in the view. Maybe this is a question for somebody else--maybe @cmungall. From my perspective, every ID (discounting the fact that their labels collide) represented in that list seems to referring to a distinct entity, so I'm not quite sure what the expected behaviour should be (again, ignoring the not-so-great labelling).
• Column 3 is not a download choice. You are trying to find GAF column 3 in the download, however, that would currently only be available in the "annotations" view, not the "gene product/bioentity" view. If you're trying to work with the data as if it was a view on the GAFs, you'd probably want to be working with the former view. If that particular information is desired for this view, it will need to be added separately at load time.

I'm going to make this issue about the first. This is touched on in berkeleybop/bbop-js#14, so I'm tempted to dupe this or the other. I've changed this one to reflect a more broad labelling issue (and it has better examples).

[cmungall]

@kltm, let's not close this ticket prematurely - afraid berkeleybop/bbop-js#14 is written in language I think only you understand...if you like perhaps start a separate ticket that is specifically about the naming of what we internally call the bioentity. Whatever we call this, it should be consistent across the board. "Acc" is weird (and probably my fault).

The third second issue is a golr loading bug and could be assigned to @hdietze -- gene_association.goa_human uses SAA1 for P0DJI8, whereas for whatever reason it is ending up as SAA1_HUMAN in golr. See:
http://amigo.geneontology.org/amigo/gene_product/UniProtKB:P0DJI8
You are right though, it has nothing to do with column labels.

Not fully sure I understand your third point. But I think making the column labels consistent should help clarify this.

[kltm]

Thinking about the implementation for berkeleybop/bbop-js#14 and this, they are distinct issues and not dupes (edited above).

@cmungall , I think you meant to say "second" above, edited for clarity.

I've added the ticket #196 and will edit this item to reflect just the scope of bad general labelling.

[kltm]

Not sure if dupe now with #195.