Update chemical probes to Probes&Drugs v01.2022

[ireneisdoomed]

P&Ds has made a new release and we should therefore rerun the current chemical probes pipeline.

Changes

Based on their release notes:

• Updated to ChEMBL30
• Updated all of the high-quality sources of chemical probes. The new version of the High-quality set (HQCP), currently with 669 compounds (+49 compared to the previous version)
• 3 brand new compound sets:
  • Natural product-based probes and drugs. A selection of 500 chemical probes recently discovered from diversity-oriented synthesis libraries
  • JUMP-Target 1 and JUMP-Target 2. 384-well plate maps of 306 compounds and corresponding genetic perturbations, designed to assess connectivity in profiling assays.

[ireneisdoomed]

Probes have been regenerated based on the data in the 01.2022 SQL dump. Apart from the changes mentioned above, we are also bringing 2 more probes sets based on their list

• Chemical Probes for Understudied Kinases
• Chemical Probes.org in favour of Chemical Probes.org (legacy) to adapt to their new portal

However data are not uploaded yet because there is a parsing issue. This a breakdown of the representation of the probes per data source:		22.02	22.04
Open Science Probes	112	130
Probe Miner	3256	3255
Probes & Drugs Portal	93	0
High-quality chemical probes	0	940
opnMe Portal	73	87
Chemical Probes.org (legacy)	602	0
Chemical Probes.org	0	748
Bromodomains chemical toolbox	57	55
Gray Laboratory Probes	88	134
Protein methyltransferases chemical toolbox	28	28
Nature Chemical Biology Probes	51	51
SGC Probes	162	170
JUMP-Target 2 Compound Set	0	72
JUMP-Target 1 Compound Set	0	72
Natural product-based probes and drugs	0	19
Chemical Probes for Understudied Kinases	0	41

“High Quality probes” shouldn’t be considered a source, as these are reported with the flag isQuality. Also “Probes & Drugs Portal” is not represented due to the same reason: these are probes in the high quality set but which are not present in other sources. I’ll get back to this bug right after Easter.

WIP notebook can be found at: https://github.com/opentargets/evidence_datasource_parsers/blob/il-22.04/exploration/chemicalProbes/chemical_probes.ipynb

[ireneisdoomed]

The module has been written and we've been able to generate a dataset based on the P&D newest version without the above mentioned parsing bug.

However there is a bug in the data itself as reported by the ChemicalProbes.org people, which hasn't been solved yet.

I will open the PR with the code as it is now, and we can tackle the data bug in a follow-up ticket.