API allows data source weights over 1

[ireneisdoomed]

Describe the bug When adjusting the data source weights in the API, it is possible to set a weight over 1. This contradicts the documentation and the calibration purpose of the weight.

Observed behaviour The API allows for weights to be adjusted over the value of 1, leading to weird behaviours. Using this query as an example:

• When the weight is 1 or 0.5, score is 0.03 and it is at the bottom. As expected, data score does not change.
• When the weight is 100 or 50, score changes to 0.39. As explained by @d0choa, this type of behaviour is unpredicted.

  query DiseaseAssociationsQuery {
  disease(efoId: "EFO_0005774") {
  id
  associatedTargets(
    page: { index: 0, size: 1 }
    orderByScore: "score"
    BFilter: ""
    aggregationFilters: []
    datasources: [{ id: "europepmc", weight: 1.0, propagate: false }]
  ) {
    rows {
      target {
        approvedName
      }
      score
      datasourceScores {
        id
        score
      }
    }
  }
  }
  }

Expected behaviour Weights should be constrained between 0 and 1 to ensure they are used as calibration factors, consistent with the documentation.

[prashantuniyal02]

@ireneisdoomed does this issue still exist? @remo87 is there any progress on this?

[ireneisdoomed]

It does. Weights > 1 change the source scoring, and the amount of evidence returned.

# response of the above query where EPMC's weight is 1
{
  "data": {
    "disease": {
      "id": "EFO_0005774",
      "associatedTargets": {
        "rows": [
          {
            "target": {
              "approvedName": "sodium voltage-gated channel alpha subunit 1"
            },
            "score": 0.9239646527845414,
            "datasourceScores": [
              {
                "id": "chembl",
                "score": 0.9992036227065824
              },
              {
                "id": "uniprot_variants",
                "score": 0.9837955517830581
              },
              {
                "id": "genomics_england",
                "score": 0.9671529481664186
              },
              {
                "id": "eva",
                "score": 0.9493026004383566
              },
              {
                "id": "orphanet",
                "score": 0.8897742701710087
              },
              {
                "id": "clingen",
                "score": 0.8464592008411808
              },
              {
                "id": "uniprot_literature",
                "score": 0.8274613634158176
              },
              {
                "id": "impc",
                "score": 0.8066729120208054
              },
              {
                "id": "ot_genetics_portal",
                "score": 0.7703459968314524
              },
              {
                "id": "eva_somatic",
                "score": 0.547137717850459
              },
              {
                "id": "expression_atlas",
                "score": 0.3393346078187555
              },
              {
                "id": "gene_burden",
                "score": 0.1899783742536316
              },
              {
                "id": "europepmc",
                "score": 0.034787151196664985
              }
            ]
          }
        ]
      }
    }
  }
}

# response of the above query where EPMC's weight is 10
{
  "data": {
    "disease": {
      "id": "EFO_0005774",
      "associatedTargets": {
        "rows": [
          {
            "target": {
              "approvedName": "heparan sulfate proteoglycan 2"
            },
            "score": 2.5021322582771908,
            "datasourceScores": [
              {
                "id": "europepmc",
                "score": 0.3988921041562052
              },
              {
                "id": "impc",
                "score": 0.5075861145621914
              }
            ]
          }
        ]
      }
    }
  }
}