[dweindl]

The GitHub Pages landing page now has sortable and filterable overview tables for the PEtab v2 and v1 encodings, replacing the bokeh table. The README table is reduced to version-independent information; per-problem statistics are only on the site. Detailed per-problem pages will follow separately.

Preview (built from this branch, deployed from my fork): https://dweindl.github.io/Benchmark-Models-PEtab/

Site (internal package benchmark_models_petab/_site/, built by bmp-create-overview --html-file=... as before, so gh-pages.yml is unchanged):

• Tabulator tables (CDN) rendered from a Jinja2 template; the site extra now requires jinja2 instead of bokeh.
• Problem descriptions from metadata.yaml; per-column filters (substring for text; 10, >10, <=5, 10-100 for numbers; tri-state for booleans).
• Summary tiles (problem counts per PEtab version; ranges of simulation conditions/experiments, estimated parameters, measurements, observables over all problems), column groups, tooltips, and distribution tags.
• Ctrl+C copies the visible rows including headers; CSV download per table; light/dark toggle.

v2 statistics (overview_v2.py): experiments (instead of v1 simulation conditions), experiments with preequilibration / steady-state measurements, estimated parameters, measurements, observables, species, events, possible discontinuities (SBML-based guess, or experiments with condition changes after the start), noise and prior distributions as named in the v2 files, references, SBML4Humans links. Problems with multiple SBML models are supported. v1.MODELS/v2.MODELS list the problems with a v1/v2 encoding.

README: the table generated by bmp-create-overview --update now has the columns ID, PEtab v1, PEtab v2 (links to the available encodings), and Description, plus a short note on the variety of problems with a link to the site for the statistics. bmp-create-overview --markdown prints this table; without options, it prints the v1 statistics.

🤖 Generated with Claude Code

[dilpath]

Thanks! Did not review in great detail -- let me know if you want feedback on something specific. Overall, the new website looks very useful.

[dilpath]

Users might be confused because there is no "state variables" column in the github.io page.

[dweindl]

Ah, yeah, I'll remove it here. I had it there, but removed it again. Interesting in principle, but the problem is that this is pretty implementation-dependent. And some proxy such as number of SBML species is potentially confusing, because some models have zero, because they are only defined in terms of RateRules targetting parameters. Can be added later on if we settle on a meaningful quantity.

[dweindl]

Did not review in great detail -- let me know if you want feedback on something specific.

That's fine. I am planning to add further content later on. It's mostly about whether you are okay with restricting the README table to high-level info and moving the more detailed records to GitHub pages.

Overall, the new website looks very useful.

Great, thanks.

[dilpath]

Did not review in great detail -- let me know if you want feedback on something specific.

That's fine. I am planning to add further content later on. It's mostly about whether you are okay with restricting the README table to high-level info and moving the more detailed records to GitHub pages.

Ah right. Maybe the most important thing taken out of the README are the references. I think the "Links" columns could be added to the README, and moved to the left of the table on the webpage -- I would say I have most frequently used the table for the links. But not necessary.

[dweindl]

Maybe the most important thing taken out of the README are the references. I think the "Links" columns could be added to the README,

Makes sense, but let's do that separately: Add the key publication to the version-independent metadata file and include it from there.

and moved to the left of the table on the webpage -- I would say I have most frequently used the table for the links.

Same here. Changed.