Update Chemical Probes summary widget and detail view

[andrewhercules]

See comment below for full design specification — https://github.com/opentargets/platform/issues/1677#issuecomment-896677359

[JarrodBaker]

Steps required to unblock from BE:

• [x] Update PIS to transfer chemical probes raw data file
• [x] Incorporate new dataset into ETL
• [x] Expose data via API

[andrewhercules]

With the new target ETL pipeline and API response, we have new Chemical Probe insights that we can display on the target profile page.

query targetInfo {
  target(ensemblId: "ENSG00000204256") {
    id
    approvedSymbol
    chemicalProbes{
      id
      control
      drugId
      inchiKey
      mechanismOfAction
      origin
      probeMinerScore
      probesDrugScore
      scoreInCells
      scoreInOrganisms
      targetFromSourceId
      urls{
        niceName
        url
      }
    }
  }
}

Sample query: BRD2 (ENSG00000204256)

Can we please update the Chemical Probes summary widget and detail view to show the new data?

Summary widget

In the current production Platform, we display the sources of the chemical probes data, separated by a bullet:

Instead, can we please show the total number of chemical probes returned in the chemicalProbes array?

Detail view

The new Chemical Probes detail view will only have one data table, showing data for all different sources.

• [ ] Please remove the section and logic that shows the "Potential chemical probes can be explored with ProbeMiner." string

-[ ] Please remove the old table component (OtTableRF) and replace with the new Open Targets table component

• [ ] Please create a new data table based on the design specification available on the team Google drive:

Probe ID	Reported target	Mechanism of action	Probe origin	Score	Source
id with control as tooltip, if available	targetFromSourceId	mechanismOfAction	origin	Score chips (see below)	n sources where n = length of urls array

• [ ] In the "Probe ID" column, please link the id to the drug profile page if drugId is not null. Also, if control is not null, please display a tooltip with the following string — "Control: control"

• [ ] In the "Reported target" column, please link to UniProt using identifiers.org - https://identifiers.org/uniprot: + targetFromSourceId.

• [ ] In the "Score" column, please display chips in Open Targets blue #3489ca for each of the scores and their respective value, if not null

Probe Miner: probeMinerScore Drugs: probesDrugScore Cells: scoreInCells Organisms: scoreInOrganisms

For each chip, please show an on-hover tooltip with the score and the description of the score that has been provided by @ireneisdoomed below (see https://github.com/opentargets/platform/issues/1677#issuecomment-896893979).

The overall style and interaction pattern for the chips can copy the approach taken for the driver method chips in the intOGen evidence data table.

• [ ] In the "Source" column, please use the publications drawer component and show the niceName of each element in the urls array. Please also link the niceName using the element's url value, if available.

• [ ] To test the final implementation, please use targets like BRD2 (ENSG00000204256), LIMK1 (ENSG00000106683), PIK3CA (ENSG00000121879), and MAPK14 (ENSG00000112062).

[d0choa]

Looks great. Only one minor comment. The score columns: Probe Miner, Drug, Cell and Organism have a lot of nulls. Each row very rarely have more than one non-null and often they are all nulls. I feel they might do the table unnecessarily wide.

An alternative could be to have a single column with the header "Scores" and have chips or other element that say Probeminer: 20 Drug Score: 70, etc. Sort of what we did for intogen.

We should also have on-hover interactions to explain what each score means. @ireneisdoomed can you come out with labels for each of the 4?

On a different note, the control column is very strange. It can have names DDR-TRK-1N but also integers 1. It's secondary information and also very frequently null. We could make it a tooltip of Probe ID that says Control: DDR-TRK-1N.

[andrewhercules]

Good idea @d0choa! I have created a design mockup and updated the ticket

[ireneisdoomed]

The mock-ups look awesome @andrewhercules!

Proposed labels for the chips:

• Probe Miner (probeMinerScore): "The Probe Miner score is the sum of 6 parameters: target potency, target selectivity, cell potency, structure-activity relationships (SAR), inactive analogs and pan-assay interference (PAINS)."
• Probes&Drugs (probesDrugScore): "The P&D probe-likeness score is the sum of 6 parameters: target potency, target selectivity, cell potency, potency-selectivity synergy, presence of a control compound and presence of an orthogonal probe."
• Cells (scoreInCells): "The organism score ranks the chemical probe for its use in model cells."
• Organisms (scoreInOrganisms): "The organism score ranks the chemical probe for its use in model organisms."

[d0choa]

Looks great @LucaFumis. After discussion with @ireneisdoomed, we have a few minor suggestions:

• [x] Update Source in the description to Probes & Drugs as specified by @andrewhercules in mock design.

• [x] Remove Reported target column. After exploring the data, it does look highly redundant and it adds very little information.

• [x] On the position of the removed column, please add another column titled Quality. Using the output of the isHighQuality endpoint, add a star (like the one used in clinvar evidence) colored in primaryColor if true or with the alternative style if false. Also include a tooltip next to the Quality column name with the following description: "Chemical probes selection based on the union of following criteria: compound belongs to one of the high-quality probe sets; use in Cells or Organisms rating ≥ 75%; P&D approved experimental probe; not labelled as obsolete."

• [x] We would like to reorder the rows so that all isHighQuality == true are on top and isHighQuality == false records appear at the bottom of the table. We might need a second layer of ordering if still not clear but let's see how it looks in a first pass.

@ireneisdoomed is looking at how to make the URLs more specific/useful from the data side and also removing the label (suitable probes) from the Probe Miner reference niceNames. None of these changes will impact FE

[LucaFumis]

Front end work done.