Closed deniseOme closed 5 years ago
I have checked the numbers of each of the different categories in the evidence data from the 19.02 release and this is what I found:
Counts based on the target.activity field in the JSON
Target activity | Counts |
---|---|
drug_negative_modulator | 391178 |
drug_positive_modulator | 63493 |
up_or_down | 22293 |
Counts based on the _actiontype field in the JSON
Target activity | Action type | Counts |
---|---|---|
NEGATIVE MODULATOR | RELEASING AGENT | 340 |
NEGATIVE MODULATOR | ANTAGONIST | 27840 |
NEGATIVE MODULATOR | INVERSE AGONIST | 347 |
NEGATIVE MODULATOR | ANTISENSE INHIBITOR | 0 |
NEGATIVE MODULATOR | ALLOSTERIC ANTAGONIST | 192 |
NEGATIVE MODULATOR | NEGATIVE ALLOSTERIC MODULATOR | 3310 |
NEGATIVE MODULATOR | INHIBITOR | 332247 |
NEGATIVE MODULATOR | DEGRADER | 1 |
NEGATIVE MODULATOR | NEGATIVE MODULATOR | 10 |
NEGATIVE MODULATOR | BLOCKER | 10 |
OTHER | SUBSTRATE | 20 |
OTHER | MODULATOR | 18635 |
OTHER | CHELATING AGENT | 0 |
OTHER | HYDROLYTIC ENZYME | 1521 |
OTHER | DISRUPTING AGENT | 135 |
OTHER | SEQUESTERING AGENT | 0 |
OTHER | METHYLATING AGENT | 0 |
OTHER | STABILISER | 750 |
OTHER | REDUCING AGENT | 0 |
OTHER | OTHER | 340 |
OTHER | PROTEOLYTIC ENZYME | 16 |
OTHER | BINDING AGENT | 538 |
OTHER | OXIDATIVE ENZYME | 0 |
OTHER | CROSS-LINKING AGENT | 336 |
POSITIVE MODULATOR | OPENER | 1701 |
POSITIVE MODULATOR | POSITIVE MODULATOR | 4816 |
POSITIVE MODULATOR | ACTIVATOR | 2108 |
POSITIVE MODULATOR | AGONIST | 35341 |
POSITIVE MODULATOR | POSITIVE ALLOSTERIC MODULATOR | 18135 |
POSITIVE MODULATOR | PARTIAL AGONIST | 1392 |
For me the biggest question is why we seem to use "antagonist", "agonist" and "up_or_down" as activities in the platform, which is a mixture of the activity and action data we get from ChEMBL.
I will investigate this with Paula from ChEMBL as part of an effort to understand all fields in the ChEMBL evidence strings.
Update: It looks like ChEMBL does the mapping of different activity types to drug_positive_modulator, drug_negative_modulator and up_or_down because they provide these terms in the target.activity
field and they also provide their activity types in the target2drug.action_type
field.
The target.activity
field entries are changed to agonist, antagonist and up_or_down in the front end.
It makes sense for us to use the terms in the action_type field to be more specific. This information has been added to this document which specifies upcoming changes to the JSON schema for ChEMBL evidence strings.
For the 19.09 release, ChEMBL have changed the activity
values to be negative_modulator
, positive_modulator
and other
. The plan is to display both, these general categories and ChEMBL's more specific terms in the drug tables.
This ticket can be closed since details for the front end have been summarised in ticket #730.
Following a support email "I totally understand what "antagonist" and "agonist" mean in the "activity" column. However, I am not able to uncover the meaning of "up_or_down". Could you help me on that? Where could I find this information?"
According to @gkos-bio "It means mainly that this was a temporary solution until we could better qualify the MOA of the small molecule or biologic. Maybe, it’s time to revisit the classification of MOAs in Open Targets. This will require more work from and coordination with ChEMBL."
From Paula (ChEMBL): _up_or_down seems to be one of the activity types in the JSON schema:
"activity": { "type": "string", "enum": [ "http://identifiers.org/cttv.activity/decreased_transcript_level", "http://identifiers.org/cttv.activity/decreased_translational_product_level", "http://identifiers.org/cttv.activity/drug_negative_modulator", "http://identifiers.org/cttv.activity/drug_positive_modulator", "http://identifiers.org/cttv.activity/gain_of_function", "http://identifiers.org/cttv.activity/increased_transcript_level", "http://identifiers.org/cttv.activity/increased_translational_product_level", "http://identifiers.org/cttv.activity/loss_of_function", "http://identifiers.org/cttv.activity/partial_loss_of_function", "http://identifiers.org/cttv.activity/up_or_down", "http://identifiers.org/cttv.activity/up", "http://identifiers.org/cttv.activity/down", "http://identifiers.org/cttv.activity/tolerated", "http://identifiers.org/cttv.activity/predicted", "http://identifiers.org/cttv.activity/damaging", "http://identifiers.org/cttv.activity/damaging_to_target", "http://identifiers.org/cttv.activity/predicted_tolerated", "http://identifiers.org/cttv.activity/predicted_damaging", "http://identifiers.org/cttv.activity/tolerated_by_target", "http://identifiers.org/cttv.activity/unknown" ], "description": "Activity of target in disease context" In our data we assign 3 types of activities:
drug_negative_modulator for drugs which are negative modulators, drug_positive_modulator for drugs which are positive modulators, up_or_down for any other
So I think this must have been how it was required (from many years ago I think)._
This is how ChEMBL classifies the different activity types (what they seem to provide to us is the parent term i.e. negative modulator, positive modulator. But do we have "other". Or do we use "up_or_down" instead of "other"):
Things we may need to do:
1) revisit our classification of activity
Considerations: Where is this activity data in the ChEMBL website? Is it under "activity charts"? If so, there does not seem to be a correspondence between what we have ("up_or_down" for example) and what they provide. Ex:
--> ChEMBL website: no data available: https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL2108576/
--> ChEMBL website: other, plus hepatotoxicity https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL1201550/ (if we click on "other" in the bioactivity summary chart we get redirected to: https://www.ebi.ac.uk/chembl/g/#browse/activities/filter/molecule_chembl_id%3A(%22CHEMBL1201550%22)%20AND%20standard_type%3A(%22Hepatotoxicity%20(cytolytic)%22%20OR%20%22Hepatotoxicity%20(granulomatous%20hepatitis)%22%20OR%20%22Hepatotoxicity%20(malignant%20tumour)%22%20OR%20%22Hepatotoxicity%20(mechanism)%22%20OR%20%22Hepatotoxicity%20(severe%20hepatitis)%22%20OR%20%22Hepatotoxicity%20(steatosis)%22%20OR%20%22Hepatotoxicity%20(successful%20reintroduction)%22%20OR%20%22Hepatotoxicity%20(time%20to%20onset)%22)
ChEMBL website: Inhibition... https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL1655/
--> ChEMBL website: no data available: https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL1201649/
@AsierGonzalez says "I guess it would be interesting to know how many drugs there are for each activity category. Perhaps there are only a handful of “up_or_down” cases.