Missing parent: [GO term] histone H3K36 methyltransferase activity

[ValWood]

• GO term ID and label for which you request a new superclass

GO:0046975 | histone H3K36 methyltransferase activity

• New superclass (parent) suggested GO:0018024 | histone lysine N-methyltransferase activity

according to the defs

[hdrabkin]

Also added this parent to the other H3Knumber activities (K27, K37, K4, K56, K79, andK9) Done

[pgaudet]

Hi,

just wondering of this is a good axis of classification - with @colinlog we wanted to remove the classification by amino acid, to only keep by histones. Specifically, we have two axes of classification for 'histone methyltransferase activity':

by residue: 'histone arginine N-methyltransferase activity' 'histone glutamine N-methyltransferase activity' 'histone lysine N-methyltransferase activity'

by histone: 'histone H2A methyltransferase activity' 'histone H3 methyltransferase activity' 'histone H4 methyltransferase activity'

We wanted to remove the classification by residue. Is that OK for you?

(Note that we have the same issue for kinase activity; not all modifications have these 2 axes)

Thanks, Pascale

[ValWood]

Then methyltransferases would be different from all of the other modifiers, and we would lose specificity

e.g. SPAC29B12.02c | set2 | histone lysine H3-K36 methyltransferase Set2 SPCC297.04c | set7 | histone lysine H3-K37 methyltransferase Set7 SPCC306.04c | set1 | histone lysine H3-K4 methyltransferase Set1 SPBC428.08c | clr4 | histone lysine H3-K9 methyltransferase Clr4 SPCC4B3.12 | set9 | histone lysine H4-K20 methyltransferase Set9

This would make modelling more difficult. I thought we decided years ago that we would model histones in a residue-specific way so that we could capture the histone code. I and the specific modified residues are important for the recruitment of effectors of specific downstream pathways. I don't know how I would do this otherwise.

What is the rationale?

[hdrabkin]

@ValWood , I suppose you could get a PRO id for each specific proteoform and position and then use 'has output' for the activity annotation (say histone lysine methyltransferase activity has output specific id for H3-K9. Just a thought

[ValWood]

I don't have a way to do has_output, and this would be different from the way we have been doing all of the other histone modifications for years.

We already request PRO Ids for active forms and for substrates. For example,

for example

[histone H3K9 methyltransferase activity] clr4 has substrate hht1 (InitMet-/Ubiq:(K14)) involved in pericentric heterochromatin formation

(here the input to clf4 H3K9 methyltransferase is a hht already ubiquitinated on K14)

Nobody uses has_output in extensions only has_input. This would add another level of complication because we already need to think about the active (modified) form of the entity itself and the modified form of the substrate without adding the output of the substrate (which would need to differentiated from the modified form of the input).

It seems to be making everything unnecessarily complicated, especially when the current system works really well.

[ValWood]

Q how would the users find/query the histone lysine H3K37 methyltransferase or the histone H3K4 acetyltransferase activity

We already did the big clean up of removing the equivalent process terms, but these MF specificities are really useful for linking MF to the different specific BPs that they regulate or are part of.

[ValWood]

Out of date. Amino acid classification will go. Histone and residue will stay.