stats required for sustainability issues

[ValWood]

This is related to the GBCR (and was a question at my departmental talk last week!) When will we reach sustainability (i.e will be able to add data to the database immediately as it is published).

I don't fully know but I think we would have definitely achieved it during this grant if we had 3 curators.

I'd like to see "The difference between the number of ‘curatable’ (6695) and ‘curated’ (4423) publications (2272) provides an alternative measure of curation completion (66% for PomBase)."

over time, because it is definitely going up more slowly (mainly since Antonia left), and probably now decreasing since Midori left and I began GO work. There are just too many other tasks to do for me to focus fully on curation.

This information would enable us to gauge what curator capacity would be required for sustainability.

[ValWood]

Year curatable curated (curatable, submitted date) 2011 2000 200 (10%) 2012 2500 400 (20) 2013 2800 2015 68 2023 5000 3600 (66%)

[kimrutherford]

Here's a one-off set of stats of curated vs curatable. I'm working on generating these numbers as part of the nightly pipeline.

Here are the numbers per year: stats-year.tsv.txt and per month: stats.tsv.txt

date	curated	curatable	uncurated	percent_curated
1969	0	1	1	0.0
1970	0	2	2	0.0
1971	0	4	4	0.0
1972	0	4	4	0.0
1973	0	11	11	0.0
1974	0	18	18	0.0
1975	0	22	22	0.0
1976	0	28	28	0.0
1977	0	43	43	0.0
1978	0	48	48	0.0
1979	0	55	55	0.0
1980	0	60	60	0.0
1981	0	72	72	0.0
1982	0	82	82	0.0
1983	0	89	89	0.0
1984	0	104	104	0.0
1985	0	119	119	0.0
1986	0	152	152	0.0
1987	0	182	182	0.0
1988	0	215	215	0.0
1989	0	258	258	0.0
1990	0	318	318	0.0
1991	0	395	395	0.0
1992	0	485	485	0.0
1993	0	582	582	0.0
1994	0	706	706	0.0
1995	0	843	843	0.0
1996	0	1004	1004	0.0
1997	0	1160	1160	0.0
1998	0	1395	1395	0.0
1999	0	1600	1600	0.0
2000	0	1813	1813	0.0
2001	0	2021	2021	0.0
2002	0	2253	2253	0.0
2003	0	2486	2486	0.0
2004	0	2725	2725	0.0
2005	0	2958	2958	0.0
2006	0	3159	3159	0.0
2007	0	3385	3385	0.0
2008	0	3634	3634	0.0
2009	0	3862	3862	0.0
2010	0	4086	4086	0.0
2011	0	4300	4300	0.0
2012	371	4559	4188	8.1
2013	827	4805	3978	17.2
2014	1729	5043	3314	34.3
2015	2309	5268	2959	43.8
2016	2735	5476	2741	49.9
2017	3023	5686	2663	53.2
2018	3336	5881	2545	56.7
2019	3683	6084	2401	60.5
2020	3915	6243	2328	62.7
2021	4119	6434	2315	64.0
2022	4254	6594	2340	64.5
2023	4434	6708	2274	66.1

[kimrutherford]

Percentage curated:

[kimrutherford]

[kimrutherford]

[ValWood]

This is interesting. I'm glad it isn't dropping, but you can cleary see the change from 3,2, & 1 curators

[kimrutherford]

I've started a stats page that's updated nightly: https://pombase.org/curation_stats

There are no links to it yet.

[ValWood]

We could also put the number of annotations/genes per paper on this page. because in addition to number changes with staff changes, we also need to explain the overall fewer papers curated as time goes on

[kimrutherford]

We could also put the number of annotations/genes per paper on this page.

I've started work on that: https://www.pombase.org/curation_stats

Please ignore the annotations per publication graphs for now as they broken.

The numbers plotted in the "Average annotated genes per publications by year range" graph are slightly different to the Canto stats page. I did quite a bit of head scratching over that but I think the new graph is correct and the Canto graph is slightly wrong.

For the Canto graph, genes are only counted if they have annotation in a Canto session. But there are some publications that have approved sessions with no Canto annotation. Examples:

• all the annotation comes from the legacy GO file (example: PMID:3912263 4e90eb1e64def030)
• all the annotation is from a LTP PHAF file: PMID:12618370

The new graph includes those genes. The min and max numbers are slightly more extreme in the new graph:

• 1986-1990 old: 2.36 new: 2.3 (genes per paper)
• 2016-2020 old 9.83 new: 10.3

[kimrutherford]

Please ignore the annotations per publication graphs for now as they broken.

Those graphs are much improved now. The numbers don't much the Canto graphs but they're in the right ballpark. I'll investigate the differences later.

https://www.pombase.org/curation_stats

[ValWood]

It looks good though, this will be very useful

[kimrutherford]

The 2021-2025 bucket on "Average high throughput annotations per publication by year range" graph is very large thanks to that big paper from Jurg's lab and there haven't been many other paper in that range to bring the average down. I wonder if there is a different way to present these numbers?

[ValWood]

Yes, this is a problem , If I was talking about the data I could explain it. We can just eave out the HTP graph because the thing I am trying to demo is LTP (that is where most of the effort is)

[ValWood]

Is this ticket finished?

[kimrutherford]

Yep, let's close this. When I get back to this, I'll open a new issue for new graphs.