japonicusdb / japonicus-config

Configuration for JaponicusDB
0 stars 1 forks source link

Create a new "original load japonicus slim" #79

Closed ValWood closed 3 years ago

ValWood commented 3 years ago

Create an "original load japonicus slim" pre filtering (using filtered GO terms, but not filtered mappings). (as discussed today)

kimrutherford commented 3 years ago

pre filtering (using filtered GO terms, but not filtered mappings).

I ran a load like that and put it here: http://japonicusdb-goa.kmr.nz/vis/from/id/b3c8c004-adbd-4409-972a-1dda7eff46e8

It's on my desktop so it won't be available between 1pm and 8pm or so each day.

ValWood commented 3 years ago

OK, there is still a massive increase for each aspect even if we use pre filtered for blocked mappings. It is reduced, especially for MF, but it. is still significant, and highly visible by eye. This is good because although we could exclude some of the InterPro mappings which are still false positive, I would not be able to do anything with the keywords and UniProtKB-SubCell which has a small % of false positives cover lots of entries. So, go ahead with the figure using the pre-filtered data. It will still look good.

kimrutherford commented 3 years ago

Here's how it looks with filtered GO terms, but without filtered mappings. It's changed only a bit. Please let me know which connecting lines to add or remove, or any other changes.

quilt-comp-8-export

Here's the SVG file: quilt-process-change-diagram-8.svg.gz

ValWood commented 3 years ago

brilliant! I think where we filter KW mappings there will usually be InterPro.

I'm not going to use the biggest increases, we will highlight some of the smaller , but more relevant ones.

Try BP membrane organization lipid metabolic process

CC ER mitochondria (as now)

MF hydrolase molecular adaptor activity (new)

then still connect the unknowns.

kimrutherford commented 3 years ago

Try

I'll include those changes when I recreate the figure with the change from pombase/website#1758. I'll try that on Monday.

kimrutherford commented 3 years ago

I think where we filter KW mappings there will usually be InterPro.

I don't understand that bit.

I'm not going to use the biggest increases, we will highlight some of the smaller , but more relevant ones.

Have I got this right?:

quilt-comp-11-export

ValWood commented 3 years ago

I think where we filter KW mappings there will usually be InterPro. I don't understand that bit.

I'm just pondering why the affect from using the pre-filtering data was not as large as Iexpected. You can ignore...

ValWood commented 3 years ago

Yes!

  1. We can add the connection between BP gene expression because we do mention that in the legend
  2. Remove "organelle localization" from the Quilt processes. Although it has ~175 annotations, it isn't a good slim classifier because most are also annotated to other more specific terms with higher precedence (this applies to Quilt generally)
  3. swap "chromatin remodelling". for "chromatin organization"
  4. Change the precedence of "generation of precursor metabolites and energy" so that. it comes immediately before "small molecule metabolic process". Most energy pathways will besucked up by this term so we don't see this change. Energy metabolism is important (this applies to Quilt generally)- add a connector here because this will grow (I hope) MF and CC, no changes.

This looks great!

ValWood commented 3 years ago

revised above comment

kimrutherford commented 3 years ago

Are changes 2-4 for the figure? I'd rather not have to redo it again.

ValWood commented 3 years ago

They were, but they aren't critical....I can remove the mention of energy generation from the legend. I'm probably the only person on the planet who 2&3 will mean anything to, and they don't affect the results. So jut adding back the connection for gene expression then this one is done.

kimrutherford commented 3 years ago

That's a relief! I'll add the gene expression connection a bit later this morning and post the result.

kimrutherford commented 3 years ago

How's this. Anything need tweaking? I'll export a high resolution version when we're ready to submit the manuscript.

quilt-comp-12-export

quilt-process-change-diagram-12.svg.gz

ValWood commented 3 years ago

Looks perfect to me!

mah11 commented 3 years ago

Very nice.

Genetics wants figures in PDF or EPS format (PDF is ever-so-slightly easier to use with LaTeX). Hint: when saving .pdf or .eps in Inkscape, DON'T tick "Rasterise filter effects".

kimrutherford commented 3 years ago

Cheers.

Here it is converted to PDF. It looks OK to me but could you have a look too?

japonicus-paper-quilt-diagram.pdf

ValWood commented 3 years ago

Fab!