Reproduce graphs from original data

[krlmlr]

Done by inst/getdata.R, I have trouble recreating yeast and USairports .

• The Nature publications required for yeast are not found, cannot open URL 'https://www.nature.com/nature/journal/v417/n6887/extref/nature750-s1.doc': HTTP status was '404 Not Found', new location linked at https://www.nature.com/articles/nature750#Sec8
• USairports relies on ~/Downloads/1067890998_T_T100D_SEGMENT_ALL_CARRIER.csv

@gaborcsardi: Can you help with the .csv file?

Moving forward, I propose to separate the downloads from the creation of the graphs so that we're less reliant on external data storage here.

[gaborcsardi]

I don't have that CSV any more, sorry.

[krlmlr]

Thanks. We could create that CSV from the graph, I suppose.

I'm thrilled that the other graphs seem to be recreatable without problems. We'll need to compare the results, though.

[szhorvat]

Just to understand, is it necessary to recreate the networks from the original source or are you just testing the scripts that do so?

[krlmlr]

No, it's not necessary, just good practice.

[maelle]

@krlmlr

Moving forward, I propose to separate the downloads from the creation of the graphs so that we're less reliant on external data storage here.

Do you mean the package should contain a copy of the datasets? Or we store them somewhere else?

[krlmlr]

I don't mind a copy of the raw data on GitHub.

[maelle]

so the task is to try and locate the yeast and airport data?

how do we compare the obtained graphs?

[maelle]

reg the yeast data, it probably is in https://www.nature.com/articles/nature750 but I don't have access (and even if we get access I suppose the data isn't really free to share :sweat_smile: )

[maelle]

oh wait this is the data: https://static-content.springer.com/esm/art%3A10.1038%2Fnature750/MediaObjects/41586_2002_BFnature750_MOESM2_ESM.doc (via https://www.nature.com/articles/nature750#Sec8)

[krlmlr]

Yes, yeast seems solved. The only tricky challenge is to reverse-engineer the CSV file, but I'm not sure it's worth it.

[maelle]

so what are the TODOs?

[krlmlr]

• Find a place for the raw files in this repo
• Split the getdata.R script, perhaps one unified download script, and one script per dataset?
• Make sure that the generated data is unchanged

[maelle]

Wait so we need 3 new issues, this issue is not really closed then?

[krlmlr]

One issue is fine.