search

pombase / curation

PomBase curation
7 stars 0 forks source link

evaluating regulation annotation #1278

Closed ValWood closed 3 years ago

ValWood commented 7 years ago

What can be classed as regulation?

So we have 1161 gene products annotated to biological regulation.

So far, regulation is always

  1. Within process, in which case it is one MF directly regulating another molecular function
  2. Upstream, in which case it is signalling or otherwise regulating a) gene expression, or b) catabolism/stability
  3. regulation of a biological quality

Violators below (after fixing some obvious ones and removing those with no regulation annotation on PomBase 2)

Midori @mah11 cdt1 replication licensing factor Cdt1 mcb1 MCM binding protein homolog Mcb1 fml1 ATP-dependent 3' to 5' DNA helicase, FANCM ortholog Fml1 rad55 RecA family ATPase Rad55/Rhp55 dmc1 RecA family ATPase Dmc1 rlp1 RecA family ATPase Rlp1 brc1 BRCT domain protein Brc1 rdl1 RAD51D-like protein 1 (possible mf regulator?)

Val

lsd2 histone demethylase SWIRM2 (predicted)

Antonia

Approved violators, via imprinting...

clearly regulating gene expression in some way, but not by ontology term

from the same paper, hetrogeneous mix, (DNA repair) this paper crops up a lot as matrix outliers too:

extent of heterochromatin assemlby

end point of a signal transduction pathway?

replication fork protection (is a child of GO:2000104 negative regulation of DNA-dependent DNA replication) is that correct. I don't even know if this should really be a term...it's one of my old ones...

For some they could be classified as above. For some I'm not sure we can say "regulation" I'll flag some for checking, but only 62 violators currently.....

Leaving these for now:

ValWood commented 7 years ago

So essentially here, we need to know why the above are classed as regulating a process, rather than directly involved in....( I think they are mainly within process, so the assumption is that they are should usually be MF regulators)

ValWood commented 7 years ago

This might be obvious but being not required for a process doesn't automatically mean that a gene product is a regulator. It might operate redundantly, or be required for fidelity etc.

mah11 commented 7 years ago

I think what would help me more would be to understand how the violators have been identified as, well, violating. IOW, what are the criteria for "being a violator"?

I feel as though I'll need to know that to be able to tell when I've fixed one, or defined why it's an allowable exception.

ValWood commented 7 years ago

it isn't one of these:

Within process, in which case it is one MF directly regulating another molecular function Upstream, in which case it is signalling or otherwise regulating a) gene expression, or b) catabolism/stability regulation of a biological quality

So there is not indication from the annotation that it is either i) involved in signal transduction or the regulation of gene expression ii) A molecular function regulator (or regulation of molecular function) descendant annotation iii) Regulation of biological quality annotation iv) catabolism (I haven't been so fussy here about 'regulation' because it isn't always annotated as 'regulation of" but this is clearly a common way to regulate processes....

Does that make sense?

There may be more 'types' that I have overlooked....

ValWood commented 7 years ago

I began with "biological regulation" and then subtracted "molecular function regulator" (and regulation of molecular function) "signal transduction" "regulation of gene expression" "protein catabolic process" (although this may include non regulators) "regulation of biological quality" "regulation of cell cycle transition" (because although not All annotated as signal transducers these are clearly signalling pathways)

ValWood commented 7 years ago

More thoughts. I don't think a mutant phenotype which affects the rate of a process, for a gene product which is clearly involved in a process, is enough for a "regulation" annotation. The rate is likely to be affected in the mutant of a gene involved in a process, but this does not necessarily imply that the gene product is a regulator in a normal situation. The process could be happening more slowly because it is a bit stuffed. They would need to do more to show it was a regulator (which activity it directly regulated, OR showing that wt over expression increased the rate of the process). This might be obvious.

ValWood commented 7 years ago

For some, they are clearly involved directly in a process, and regulating it (by regulating some molecular function) . But as yet we don't know the exact mechanism of the regulation

For example

ValWood commented 7 years ago

wpl1 should be in this list, but isn't???

mah11 commented 7 years ago

I'm gonna open a new ticket for cdt1 because it's a bit complicated. But it's been fun looking back into it.

ValWood commented 7 years ago

autophagy related

I'm never sure where autophagy begins and ends...

mah11 commented 7 years ago

Old annotation in Artemis, now deleted. Might be only phenotypes when the paper gets done in Canto.

mah11 commented 7 years ago

mcb1 GO:1903463 PMID:21813639 IMP Mcb1's actions aren't yet well characterized at the molecular level. There are plenty of phenotype annotations from this paper, so I could easily delete the GO BP. But it doesn't have any others - @ValWood, are you happy with it not having any BP (or MF) annotations?

ValWood commented 7 years ago

So why not just "DNA replication"? instead of "regulation"?

mah11 commented 7 years ago

It's been a while, but probably because the authors interpreted the phenotypes to mean that Mcb1 acts via some effect on the MCM complex, rather than more directly in replication. But it wouldn't be a big deal to change it if that's what you prefer.

ValWood commented 7 years ago

so again this will depend if pre-initiation is part of initiation?

mah11 commented 7 years ago

For mcb1 it's even broader than that, because its role hasn't been pinned down as precisely as initiation, elongation, or both.

mah11 commented 7 years ago

aaaanyhow, plodding along ...

swi3 GO:0071515 PMID:15059961 IMP GO:0043111 PMID:15367656 IC GO:0048478 PMID:15367656 IDA

From a quick overview, it looks correct for swi3 to be annotated to GO:0043111 and GO:0048478. It inherits biological regulation annotation due to paths in GO for those two terms, i.e. it's not annotated directly to anything with "regulation" in its name. There are some other problems with the current swi3 annotation (e.g. a TAS cites the wrong paper), which I think 'n hope will be sorted by whoever tackles it properly (probably won't be me any time soon), but they don't make a big difference to the regulation annotation. As noted, that comes from the ancestry of terms for processes swi3 is directly involved in.

GO:0043111 ! replication fork arrest GO:0048478 ! replication fork protection

ValWood commented 7 years ago

Yep Its the replication fork protection parentage. I dislike that term as a process ( know I created it...see my comment for nse5 and 6 complex above).... I will add swi3 to this list.

mah11 commented 7 years ago

Replication fork arrest also has is_a GO:2000104 ! negative regulation of DNA-dependent DNA replication ... but I can see the rationale for that, since arrest does make replication stop for a while. Wonder if it should move down to regulation of elongation?

Antonialock commented 7 years ago

I think mei2 is a regulatory protein, though it is not 100% clear what it does, it seems to play a role at multiple steps

"Hence, the key molecular function of the Mei2 dot in fission yeast is most likely to be the shutdown of the DSR-Mmi1 system during meiotic prophase" " the evidence indicates that Mei2 plays a key role in sequestering and inactivating Mmi1, an RNA-binding protein crucial for the removal of hazardous meiosis-specific mRNAs during the mitotic cell cycle." (so it negatively regulates the negative regulators?) (https://www.jstage.jst.go.jp/article/pjab/86/8/86_8_788/_pdf)

I think it also played a role for premeiotic synthesis, and for G1 arrest

"It remains puzzling, however, how Mei2 participates in the earlier steps of sexual differentiation, i.e. G1 arrest and mating, even though it is not absolutely necessary for these steps. One possible explanation has emerged from our recent study; activated Mei2 was found to enhance the expression of ste11 through a positive-feedback loop involving the stress-responsive MAP kinase pathway (Sukegawa et al., 2011). Ste11 is a major transcription factor controlling the switch from cellular proliferation to sexual differentiation (Sugimoto et al., 1991; Otsubo and Yamamoto, 2012). Thus, lack of the full expression of ste11 in the absence of Mei2 function might result in lower mating efficiency.

ValWood commented 7 years ago

OK for mei2...seems likely some way regulating gene expression. Just make a note and tick this one...

mah11 commented 7 years ago

rad55, dmc1, rlp1, rdl1

rad55   GO:0010520      PMID:25414342   IMP
rad55   GO:0010520      PMID:25414342   IGI     PomBase:SPCC4G3.05c (mus81)

dmc1    GO:0010520      PMID:25414342   IMP
dmc1    GO:0010520      PMID:25414342   IGI     PomBase:SPAC3C7.03c (rad55)
dmc1    GO:0010520      PMID:25414342   IGI     PomBase:SPBC28F2.07 (sfr1)

rlp1    GO:0010520      PMID:25414342   IMP
rlp1    GO:0010520      PMID:25414342   IGI     PomBase:SPCC4G3.05c (mus81)

rdl1    GO:0010520      PMID:25414342   IMP
rdl1    GO:0010520      PMID:25414342   IGI     PomBase:SPCC4G3.05c (mus81)

We discussed the GO:0010520 annotations from PMID:25414342 around when the curation was done, and decided to keep them. Unless you've changed your mind since then I'll leave them in. This is the email exchange (first bit is from me, then it alternates between Val & me):

I wonder whether some of the genes in this session could be annotated to regulation of meiotic recombination* based on the phenotypes, to reflect the effect on whether you get crossovers and/or gene conversion. I'm not sure, because even if the mechanism is understood I haven't got my head round it, and actually it sounds as though the authors aren't sure about mechanism either.

I also don't know which part(s) of meiotic recombination are affected - strand invasion? joint molecule formation? DSB processing? some or all of the above? just looking at part_of GO:0007131 in GO, and it's not making my head spin any slower.

Would you add the high-ish level reg of meiotic recomb if you were doing this approval?

  • i.e. GO:0010520 regulation of reciprocal meiotic recombination

I think I would have added "reciprocal meiotic recombination" as I'm not sure how to judge regulation vs directly involved in vs both from just genetic interactions without knowing more about the pathway.

But, I think you are suggesting "regulation" because these interactions somehow dictate the choice of pathway?

Exactly. If I'm interpreting correctly, the effects of the mutations on which pathway gets used suggests the gene products may be involved in regulating that pathway choice (and that wouldn't rule out some or all of them also having a direct role in recombination). But I'm not sure they're sure how it works, so I'm not sure whether to make any annotation, let alone recomb vs. regulation.

I think I could go with either.

I could put in the regulation term, and explicitly ask Alexander if it looks sensible.

(Alexander Lorenz subsequently reviewed the session and had no problem with the GO:0010520 annotations. It's https://curation.pombase.org/pombe/curs/903cb37e16a1d237/ro/)

mah11 commented 7 years ago

fml1

GO:0045128      PMID:22723423   IMP

https://curation.pombase.org/pombe/curs/37ebf9ba3b26613d/ro/

I don't know this stuff well at all. It looks like there are fml1 phenotypes consistent with it doing "regulation" in the same sense as for the rad55 etc. annotations - in meiotic recombination, whether breaks are repaired in a crossover or non-crossover way. I'm really guessing now: maybe the curator (Matthew Whitby) chose negative regulation of recombination because there's more crossing over in fml1delta and less with overexpression. Fml1 unwinds D loops and to the extent that I can follow the model in the last figure, that appears to be how it influences how many breaks are repaired with crossovers. But it all does my head in.

mah11 commented 7 years ago

ssa2, mas5

I happened to notice that these two had been added to my list in the original summary since I last looked at it on Friday. I strongly recommend that you NOT make any further additions in the main summary or existing comments, because we don't get email notification for edits. Use new comments instead.

Anyway, on to annotations ...

ssa2            GO:1900035      PMID:24146635   IMP
mas5            GO:1900035      PMID:24146635   IMP

These were made by community curator Aleksandar Vjestica in https://curation.pombase.org/pombe/curs/c29f51c4cc86096d/ro/.

I haven't gone back through the paper in as much depth as when I reviewed the session, but from briefly revisiting it, the Ssa2/Mas5 experiments show phenotypes only, so I don't think there's enough detail to identify an MF regulator (even for the "high Hsf1 activity" claim, it's inferred from phenotypes). We could perhaps say regulation of transcription if you're comfortable inferring that much from these phenotypes:

mas5

mas5Δ   FYPO:0002934    altered mRNA level during vegetative growth Transcript expression level evidence        Global gene expression profile (RNAseq) of deletion similar to that of heat-stressed wild type. 
mas5Δ   FYPO:0003152    increased protein level during cellular response to heat    Microscopy          assayed_using hsp90 | 
mas5Δ   FYPO:0003151    decreased protein level during cellular response to heat    Microscopy          assayed_using PomBase:gef1 | assayed_using scd1 | assayed_using rga4

ssa2

ssa2Δ   FYPO:0002934    altered mRNA level during vegetative growth Transcript expression level evidence        Global gene expression profile (RNAseq) of deletion similar to that of heat-stressed wild type.
ssa2Δ   FYPO:0003152    increased protein level during cellular response to heat    Microscopy          assayed_using hsp90 | assayed_using psi1
Antonialock commented 7 years ago

changed the csi1 annotation, it was whacky. Also moved csi2 up to mitotic spindle organization (could be upstream somehow but it has a different role from csi1...based on current knowledge I think it is fine...)

Antonialock commented 7 years ago

bob1, I changed 'regulation of conjugation with cellular fusion' to 'conjugation with cellular fusion'.

Bob1 is shown to physically interact with byr1, and both the deletion mutant and overexpression mutant are defective for mating. It interacts genetically with byr1 as byr1-OP rescues the mating defect of both bob1-delta and bob1-OP.

Maybe bob1 links cytoskeletal events to the changes needed for shmooing? who knows..

Antonialock commented 7 years ago

myo51 changed from regulation to being involved in the process + tweaked other terms in session.

Antonialock commented 7 years ago

Hsp9 - I think this one is ok and is part of "regulation of G2/M". This guy physically interacts with 14-3-3 and cdc25 and negatively regulates the activity of cdc25. Although it is there during normal growth it only seems to be important when the cells are stressed (to slow down entry into Mphase)

All in all it looks like hsp negatively regulates cdc25, which is important during stress to slow down the cell cycle.

ValWood commented 7 years ago

hsp9- makes sense....and the others too

re hsp9, "physically interacts with 14-3-3 and cdc25 and negatively regulates the activity of cdc25" isn't currently annotated... this would join it up into the network...

Antonialock commented 7 years ago

I annotated hsp9 to negative regulation of protein tyrosine phosphatase activity has_regulation_target cdc25

I think that's the best I can do for now (can't say it is a direct regulator, and can't say the interaction is direct)

ValWood commented 7 years ago

don't use these terms.. they tell you nothing about the process

use "+ve regulation G2/M transition" has regulation target cdc25

its easy for a biologist to figure if it is +vely or -vely regulating cdc25... cdc25 positively regulates the transition, so if this is positively regulating the transition it must be negatively regulating cdc25 I'm not even sure I would say it was regulating the phosphatase activity, most likely its part of the sequestering, so this is probably not true anyways...

Antonialock commented 7 years ago

I really don't like "+ve regulation G2/M transition" has regulation target cdc25

the extension should be an extension of the GO term? Cdc25 is not a "G2/M transition" so it doesn't make sense to add it here? @mah11 what do you think?

I reported a similar thing for srw1, it's annotated to mitotic G1 cell cycle arrest in response to nitrogen starvation regulates cdc2

same thing here, the extension does not make sense as an extension to the GO term.

Antonialock commented 7 years ago

but yes I agree on not using neg reg of phosphatase activity...I do get it it was just a good fudging term :p

ValWood commented 7 years ago

Yes you are correct, it doesn't make sense.

Maybe the new qualifiers would work for these (acts upstream of or something)

Antonialock commented 7 years ago

Hmm, I think you mentioned before that you weren't so keen on those because where do you stop, but maybe they would be useful for (the few cases) where we want to make a connection and cannot be more precise, with the view to remove them as soon as the science is resolved to a more precise level?

ValWood commented 7 years ago

Leave the hsp9 as it is until we resolve https://github.com/pombase/canto/issues/1308 It isn't wrong, and it will just sit in the logs with the others which currently only make sense this way.

Antonialock commented 7 years ago

well actually, as you rightfully pointed out, it might not be regulating its atcivity, it could be regulating its localization ;)

Antonialock commented 7 years ago

(but I put it back for now with the comment it is a stop gap :-) )

ValWood commented 7 years ago

I'm happy with the rest staying as they are. I'll evaluate again in a few month

So do we agree (so far) with the following:

  1. Regulation is either upstream of a process or within or downstream
  2. OR it is regulation of a biological quality
  3. Upstream regulation can be signal transduction/ regulation of gene expression/catabolism or substrate availability ( some cases of substrate availablility will be independent of signal transduction)
  4. Within process regulation can be i). Direct regulation of a molecular function (activity or binding) ii) Regulation of pathway "decision" for a branched pathway within a process (e.g. within DNA repair process specific pathway choices can be regulated, or "mating type switching/imprinting) iii) Regulation of localization
    iv) Rate limiting (slowest) step usually within a metabolic pathway v) Operating within a process to negatively regulate the process whilst some step is completed or correction is made (I guess this is always going to be a class of i) eventually, but sometimes we don't know the specific MF that is regulated, but know that it is "within pathway" regulation)

(The above within process regulation examples only cover non-signal transduction, which may have additional sub-types like amplification )

Hopefully this is a reasonable broad level view of types of biological regulation So, fir future annotations to regulation check that the annotation is compliant with this list or extend the list as necessary. I'll re-evaluate violators after V63 (no point checking for V 62 as most of the changes were post update).

Note: if a process has a "within process regulation" i.e DNA recombination, we should aim to ensure that it also gets a direct annotation to the term if we know that it is also part_of the process (this will be the case for most). Examples I can think of where this is not true are where a signalling pathway controls a later stage of a process independently (like SIN for cell separation, during cytokinesis). This is because many tools seem to be showing only direct annotations by default and you need to toggle on the "regulates" relationship. We need to make sure that they continue to slim to the correct process if "regulates" is excluded.

ValWood commented 3 years ago

closing, outstanding issue in https://github.com/pombase/curation/issues/2982