T001: Frozen bioactivity dataset is not frozen

[dominiquesydow]

In T001, we thought we froze the bioactivity dataset (by checking for activity IDs in ChEMBL 27) but it seems not to work.

Version on master branch:
Number of bioactivities queried for EGFR in this notebook: 7178
Number of bioactivities after ChEMBL 27 intersection: 7178

Running this notebook today:
Number of bioactivities queried for EGFR in this notebook: 8817
Number of bioactivities after ChEMBL 27 intersection: 8031 (I would expect 7178)

@jaimergp, I am sorry to bother you with this.
Do you understand why our intersection with the chembl27_activities.npz.zip does not produce stable results?

If not: Since I do not have the time to debug this (and you probably neither), my suggestion is to remove the chembl27_activities.npz.zip freezing bit --- and instead freeze the final output dataset output_df from this notebook to ensure stable outputs in all downstream talktorials (T002-T007).

[dominiquesydow]

The numbers we used to determine uniqueness may not be identifies; thus let's go with the freezing the final output dataset to this dataset:
https://github.com/volkamerlab/teachopencadd/blob/ea055ef7b14d57eda20482c42c7b39da0fd8368a/teachopencadd/talktorials/T001_query_chembl/data/EGFR_compounds.csv

• Rename file from EGFR_compounds.csv to EGFR_compounds_ea055ef.csv (added commit hash).
• Add step to notebook where we overwrite output_df with the content of that frozen dataset (add note for users to disable that cell if they want to use the latest dataset).