search

pombase / curation

PomBase curation
7 stars 0 forks source link

Possible paper to automate noctua model (cytoskeleton, chromosome segregation) #3267

Open manulera opened 2 years ago

manulera commented 2 years ago

Following curation of PMID:24790093 I think this could be a good paper to try to automate the generation of a Noctua model. We can discuss. See the image below (maybe you even mentioned this paper for Noctua already?):

Screenshot 2022-06-01 at 18 57 18

manulera commented 2 years ago

Mentioing you @ValWood

dexink commented 2 years ago

@ValWood I would love a semi-supervised SOM model for producing these forms of curator-QC-dependent processes Prompts for unsupervised preliminary models to be checked, approved, and then submitted for finalization and algorithm update would be optimal (both highly flexible and adequately cautious).

ValWood commented 2 years ago

@ValWood I would love a semi-supervised SOM model for producing these forms of curator-QC-dependent processes

I am not sure exactly what you mean.

We can already automate pathway building (non -metabolic) based on MF, annotation targets (using annotation extensions) , and part_of connections to pathways. We have some proof of principle- we have our autogenerated networks esyN links from here https://www.pombase.org/browse-curation/fission-yeast-bp-go-slim-terms and GO have already built some reasonable GO-CAM pathways based on PomBase curation. However, we are concentrating on cell division processes and chromatin organization rather than metabolism.

Also, there is a lot of info in this particular model that will only be made accessible by literature curation (i.e which modified entities bind to which other entities where and when). I have seen enough ML approaches on far simpler problems to know that they don't currently save any curator time.

However, we do think some semi-automation might be possible to produce a metabolic model based on inputs and outputs which can be refined by curation and rules. Manu recently suggested that we automate the generation of an 'everything possible " network, and then refine. This would be quite far in the future as metabolic pathways are not a high priority for us right now.

ValWood commented 1 year ago

This is a nice example. Where is the regulation? Cut15 is constitutive (what does Paul call this), the regulation is via cdk phosphorylation of alp7

manulera commented 1 year ago

Yes, alp7 phosphorylation by CDK during mitosis increases its interaction with Cut15, leading to the accumulation of alp7 in the nucleus. Perhaps it's not the best example, as it's not an on/off switch, but a displacement of the equilibrium towards having more alp7-alp14 in the nucleus.