excessive granularity tRNA aminoacylation

[gocentral]

I have been meaning to submit this SF item for a while (since we discussed the granularity beneath mitochondrial translation on the reference genome call).

There are a number of terms which appeared to confuse the issue under here:

GO:0006418 tRNA aminoacylation for protein translation GO:0006419 alanyl-tRNA aminoacylation GO:0006420 arginyl-tRNA aminoacylation GO:0006421 asparaginyl-tRNA aminoacylation GO:0006422 aspartyl-tRNA aminoacylation GO:0006423 cysteinyl-tRNA aminoacylation GO:0006424 glutamyl-tRNA aminoacylation GO:0006425 glutaminyl-tRNA aminoacylation etc and tRNA aminoacylation for mitochondrial protein translation with the children GO:0070143 mitochondrial alanyl-tRNA aminoacylation GO:0070144 mitochondrial arginyl-tRNA aminoacylation GO:0070145 mitochondrial asparaginyl-tRNA aminoacylation GO:0070146 mitochondrial aspartyl-tRNA aminoacylation GO:0070147 mitochondrial cysteinyl-tRNA aminoacyl etc.

These terms are an example of excessive granularity. This is not because of the representatation of component specific terms in process. This is because these terms are 'single-step' and are equivalent to the tRNA synthetase function terms In fact, they are defined as being catalysed by these enzymes!

The process of coupling alanine to alanyl-tRNA in a mitochondrion, catalyzed by alanyl-tRNA synthetase. In tRNA aminoacylation, the amino acid is first activated by linkage to AMP and then transferred to either the 2'- or the 3'-hydroxyl group of the 3'-adenosine residue of the tRNA.

I propose that these terms are obsoleted, with a recommendation that the annotations are made to the appropriate function term and the appropriate translation process term.

We may need to retain a term for aminoacylation because this appears to occur independently of translation?

Summary The granularity we are discussing should allow us to represent "component specific processes", But it is not necessary to represent "process specific functions" as we have done here.

Reported by: ValWood

Original Ticket: "geneontology/ontology-requests/5979":https://sourceforge.net/p/geneontology/ontology-requests/5979

[gocentral]

I mean't "function specific processes" ?

Original comment by: ValWood

[gocentral]

I would suggest the appropriate action is merging to the parent terms, rather than obsoleting the terms. In that way annotations are preserved, and the child term's name can become a narrow synonym to the parent term.

-- Alex

Original comment by: addiehl

[gocentral]

Probably the merge would be better then. Is it necessary to display all of the existing terms names though as synonyms as there would be 40 ? Also, these terms have not been used by many so the re-annotation would be fairly trivial. This would serve as an example of a "function term obsoleted from process" which we could refer to. This may get forgotten after the merge when the history is only documented in the SF item. On balance, I suggested obsoletion as it might be helpful to deter future similar term requests, but a merge may be more appropriate.

Val

Original comment by: ValWood

[gocentral]

I originally objected to these terms as really unneeded but was overturned. However, they may still be useful as they do aid in query (since the specific aminoacid can be queried); rather than find all aminoacyl-tRNA synthetases (brings back many many) find all leucyl-tRNA syntehteases (only brings back those that aminoacylate leucine tRNA

so depends on how popular each query type is....

Original comment by: hdrabkin

[gocentral]

re GO:0006418 tRNA aminoacylation for protein translation needed because not all tRNA aminoacylation is used for protein synthesis 1.cell wall biosynthesis in certain bacteria 2.protein modification (arginylatins)

Original comment by: hdrabkin

[gocentral]

From an ontology-editing perspective, merging (using narrow synonyms) is both easier than obsoleting, and a better match for the situation. I would like to bring this item up on the GO list to give more curators a chance to comment on whether the terms are useful or clutter.

m

Original comment by: mah11

[gocentral]

Hi Harold,

Just to clarify, In response to your comment about the usefulness,

leucyl-tRNA synthetase would still be the appropriate query to bring back all the leucyl-tRNA synthetase rather than leucyl-tRNA aminoacylation i.e the ontology will be function rather than process

I am suggesting we loose the process terms leucyl-tRNA aminoacylation (which are single step) but we keep not the equivalent function terms (which are equivalent, and also already exist).

leucyl-tRNA aminoacylation would be added as a synonym of leucyl-tRNA synthetase

Val

Original comment by: ValWood

[gocentral]

Hi Val Ok; I am less adverse to this, as again, I originally thought addition was not really a good thing, especially since they are one-step reactions; thus 1 term for process should do it, although still need the split the "for translation and not for translation"; this is because Jane and I were trying a while back to work out how to relate tRNA aminoacylation to translation, and the problem was you couldn't all of the time, as some aminoacyl-tRNAs participates in things other than translation; yet, the translation community would most likely expect a relationship (any chapter on protein synthesis usually includes a paragraph or two on aminoacylation of tRNAs, since that is what they are mostly used for 99% of the time). Some, like the arginyl-tRNAs participate in both, while others like that one bacterial glycyl-tRNA is used only for cell-wall biosynthesis (Staphylococcus aureus pentaglycine peptidoglycan cross-bridge), even though it is acylated by the same enzyme that acylates the other glycine tRNA isoacceptors that do participate in cell wall biosynthesis.

Original comment by: hdrabkin

[gocentral]

Hi I agree, as aminoacylation occurs outside translation we still need this distiction. Val

Original comment by: ValWood

[gocentral]

meta-comment: it seems more appropriate to have discussions on a discussion list (where email programs support threaded discussions etc), and leave the tracker for tracking..

but anyway...

Regarding the last two comments from val and Harold: it seems we have come full circle and there is a reason for retaining the process subtypes?

Note that once we have the F->P links this reason will become moot. This is because we can make a direct part_of link between alanyl-tRNA synthetase and GO:0006418 ! tRNA aminoacylation for protein translation.

However, it sounds like it may be premature to obsolete or merge the process subtypes prior to introducing the F->P links.

I don't think this case is isolated. We need a general policy decision for other similar cases: eg subtypes of kinase activity and subtypes of phosphorylation

Original comment by: cmungall

[gocentral]

I don't think we've exactly come full circle, because Harold and Val haven't decided to keep all of the current process subtypes. They've agreed to keep the terms that distinguish tRNA aminoacylation in the context of translation from tRNA aminoacylation in other contexts (e.g. GO:0043040 tRNA aminoacylation for nonribosomal peptide biosynthetic process and GO:0006418 RNA aminoacylation for protein translation would be retained along with their parent GO:0043039 tRNA aminoacylation); they're still very interested in merging (or obsoleting) the terms that name individual amino acids.

I'm neutral, and willing to wait for MF-BP links.

m

Original comment by: mah11

[gocentral]

hi Chris No, not all of them; we can shed or merge the specific amino acids ones, but we want to keep tRNA aminoacylation % tRNA aminoacylation for nonribosomal peptide biosynthetic process %tRNA aminoacylation for protein translation [GO:0006418

and if we keep mitochondrial translation, we will need %tRNA aminoacylation for mitochondrial protein translation

Original comment by: hdrabkin

[gocentral]

Suggested action then sounds like:

(1) merge specific amino acid tRNA acylation terms into the parent term (or obsolete them) (2) keep GO:0043040 : tRNA aminoacylation for nonribosomal peptide biosynthetic process GO:0006418 : tRNA aminoacylation for protein translation GO:0070127 : tRNA aminoacylation for mitochondrial protein translation (3) link the MF terms for the individual aa synthetase activities to the appropriate BP terms (either one of those in 2 or the parent tRNA aminoacylation) using part_of

The open question is WHEN to do what:

(3) before (1) (1) before (3) (1) and (3) at the same time

If we filter out the part_of relships between MF and BP from the non go_ext files as we said we would, we can essentially have our cake and eat it too.

Original comment by: tberardini

[gocentral]

before going home I would think better to merge if there are many annotations BUT, since potentially 20 GOids to merge, will that be a problem ?

Still need to make 20 links right?

Original comment by: hdrabkin

[gocentral]

I have a slight preference for doing Tanya's (1) and (3) at the same time, and I still also slightly prefer merging to obsoleting. We have 20-odd links between MF terms and whichever BP term anyway, whether we merge or obsolete the amino acid-specific process terms, so I don't think that makes any difference. Merging is a bit easier for editors, and means annotations can be converted from alt_ids to primary IDS pretty much automatically.

The uniprot file has thousands of annotations, but they're almost all IEA.

m

Original comment by: mah11

[gocentral]

It might seem in the scheme of things that this request is not really warranted, so i'd like to make it slightly clearer why I requested this

1. This is a little bit like the a)phophorylation/ b) kinase and a) methylation/ b) methylase examples BUT in both of these cases there is complex upstream regulation and so a and b are not equivalent. I don't know of any other case in the current ontology where the function and process terms are absolutely equivalent. In other cases the process term is still required and will be related to the function term by some part-of/has_part relationship. In this case (as far as we know) these are the same. These particular terms were used as an example in the P-C discussion but this was a bad example as the terms were not required in the first place.

2. Shortly I would like to request a term for 'cytoplasmic translation'. This will need the appropriate children so it would be good to deal with these 20 surplus terms first. It is not the explosion in terms per se which is an issue, but the extra work for annotators annotating to these terms unnecessarily. it is easier to move any existing annotations back to aminoacylation and reannotions will only need to be made to a single new term rather than the 20 existing ones.

Val

Original comment by: ValWood

[gocentral]

1. I don't think anyone has argued for keeping the aminoacylation process terms that name specific amino acids. The only question has been whether to do the cleanup by obsoleting or merging, with a slight tendency to favor merging.

2. 'Cytoplasmic translation' = translation. It doesn't happen in the nucleus, or outside the cell (experiments don't count ;) ).

m

Original comment by: mah11

[gocentral]

1. OK I wasn't sure if it was clear why I had requested this as there is other potantial redundency between M&F, that this is a "specially bad case"

2. oops, cytosolic translation....

Merge is good if that is the consensus.

Original comment by: ValWood

[gocentral]

I think we're in a position to act on this now; Tanya and David have started adding some of the "obvious" links between functions and one-step processes, so it seems we could do Tanya's (1) and (3) together any time now. Emailing ...

Original comment by: mah11

[gocentral]

We need to be cautious: Cytoplasmic translation' = translation. It doesn't happen in the nucleus, or outside the cel. ) ). May not be true at all: see Journal of Cell Science 117, 5713-5720 (2004), and Differentiation. 2002 Mar;70(1):10-22. BUT: RNA. 2003 Jan;9(1):1-8 (J Dahlberg I sort of agreed with Dahlberg, but this has even now gotten into textbooks (Biochemistry, Campbell and Farrell).

if we are going to have proceses called mitochondrial translation and chloroplast translation; then we need a label for translation that isn't those two processes. However, in the cases made for nuclear translation, the processes appear to use the same components as their cytosolic counterparts (eIF's, EFs, aminoacylated tRNAs, etc.

We also need to decide if we want to distinquish prokaryotic type vs eukaryotic type translation (they are of course, similar, but there are differences in complexity (especially in the initiation process.

I'm not sure why we would want to keep process terms for the individual aminoacylations, but not function terms. The "process" is in most cases a single step (except for the cases in certain organelles and organisms where, for example, the making of gln-tRNA proceeds not by direct aminoacylation of tRNAgln with gln, but first an aminoacylation of tRNAgln with glu, then a transamidation of the glu-tRNAgln to make gln-tRNAgln.

I would feel more comformtable with function terms for each amino acid. This buys us an additional advantage that we can make the process specific functions where the aminoacylated tRNA is not destined for translation, but for cell wall biosynthesis, or post-translational protein modification.

Original comment by: hdrabkin

[gocentral]

Let's leave the location(s) of translation out of this item -- Val will open a new one when she's ready, and we can discuss which terms we need there.

Returning to tRNA aminoacylation - David and Tanya have agreed to take on this item so that the MF-BP links will be in their to-do queue. Then they'll either do the merges for the AA-specific terms or hand it back to me.

m

Original comment by: mah11

[gocentral]

• labels: 580925 --> Other term-related request
• assigned_to: nobody --> ukemi

Original comment by: mah11

[gocentral]

Hi Tanya and David,

Is this SF item on your Process/Function links to_do list?

The occurs_in XP should have taken care of the 'location of translation' issue.

Thanks! Jane, Paola and Becky

Original comment by: rebeccafoulger

[gocentral]

Yup. Still on out to-do list.

D

Original comment by: ukemi

[gocentral]

• milestone: --> PomBase

Original comment by: mah11

[ValWood]

Let's close this one. I made a not of it in the list of terms we might not need but in the grand scheme of things their existence isn't harmful.