Zenodo

GET https://zenodo.org/api/records?size=1

{
  "hits": {
    "hits": [
      {
        "created": "2026-07-05T15:36:21.675487+00:00",
        "modified": "2026-07-05T15:36:22.036973+00:00",
        "id": 21207429,
        "conceptrecid": "21207253",
        "doi": "10.5281/zenodo.21207429",
        "conceptdoi": "10.5281/zenodo.21207253",
        "doi_url": "https://doi.org/10.5281/zenodo.21207429",
        "metadata": {
          "title": "Mendelian Randomization Predicts Drug-Target Success Only for Abundance-Modulating Mechanisms: A Pre-Registered, Outcome-Blind Evaluation",
          "doi": "10.5281/zenodo.21207429",
          "publication_date": "2026-07-05",
          "description": "<p>Mendelian randomization (MR) using cis-protein quantitative trait loci (pQTLs) is widely used to prioritize drug targets, yet its predictive validity has never been evaluated with respect to the mechanism by which a drug acts on its target. We present a pre-registered, outcome-blind evaluation of a frozen, zero-parameter cis-pQTL MR classifier against 195 Phase III drug-target pairs across 24 diseases, drawn from UKB-PPP and EpiGraphDB instruments matched to Open Targets. Of these, 138 pairs (71%) had evaluable MR results; 57 were missing due to instrument-GWAS panel incompatibility.</p>\n<p>Our primary finding is a pre-registered mechanism dissociation: MR is uninformative for activity-blocking drug targets (balanced accuracy [BA] = 0.511, 90% CI [0.456, 0.576]; n = 76), confirming a priori that a genetic instrument for protein abundance cannot detect drugs that inhibit protein function without changing protein level. For abundance-modulating targets, MR trends positive (BA = 0.590, 90% CI [0.496, 0.691]; n = 59) but the confidence interval includes chance, and missingness disproportionately depletes successes from the analyzed set. A continuous ROC analysis provides the strongest pooled evidence of discrimination (AUC = 0.595, permutation p = 0.038). Leave-one-disease-out analysis confirms no single disease drives the result (BA range [0.538, 0.584]). The pooled BA of 0.559 ([0.503, 0.617]) barely clears chance and drops to 0.533 under worst-plausible missingness assumptions.</p>\n<p>The contribution is the dissociation itself: MR-based target validation has a definable domain of validity set by drug mechanism, and the boundary between informative and uninformative strata is a structural property of the genetic instrument. Within the abundance domain, a significant MR result is a positive prioritization signal (PPV = 0.56); outside it, MR is silent by construction.</p>",
          "access_right": "open",
          "creators": [
            {
              "name": "Tower, Elliot",
              "affiliation": null,
              "orcid": "0000-0001-7004-8884"
            }
          ],
          "keywords": [
            "drug target validation",
            "cis-pQTL"
          ],
          "resource_type": {
            "title": "Preprint",
            "type": "publication",
            "subtype": "preprint"
          }
        }
      }
    ],
    "total": 6861368
  },
  "aggregations": {
    "publication_date": {
      "buckets": [
        {
          "key": "2000",
          "doc_count": 11710,
          "label": "2000",
          "is_selected": false
        },
        {
          "key": "2001",
          "doc_count": 14210,
          "label": "2001",
          "is_selected": false
        }
      ],
      "label": "Publication date"
    },
    "access_status": {
      "buckets": [
        {
          "key": "open",
          "doc_count": 6619459,
          "label": "Open",
          "is_selected": false
        },
        {
          "key": "restricted",
          "doc_count": 231900,
          "label": "Restricted",
          "is_selected": false
        }
      ],
      "label": "Access status"
    },
    "resource_type": {
      "buckets": [
        {
          "key": "publication",
          "doc_count": 4482400,
          "label": "Publication",
          "is_selected": false,
          "inner": {
            "buckets": [
              {
                "key": "publication-article",
                "doc_count": 2377756,
                "label": "Journal article",
                "is_selected": false
              },
              {
                "key": "publication-taxonomictreatment",
                "doc_count": 707031,
                "label": "Taxonomic treatment",
                "is_selected": false
              }
            ]
          }
        },
        {
          "key": "image",
          "doc_count": 1221110,
          "label": "Image",
          "is_selected": false,
          "inner": {
            "buckets": [
              {
                "key": "image-figure",
                "doc_count": 834446,
                "label": "Figure",
                "is_selected": false
              },
              {
                "key": "image-photo",
                "doc_count": 363620,
                "label": "Photo",
                "is_selected": false
              }
            ]
          }
        }
      ],
      "label": "Resource types"
    },
    "subject": {
      "buckets": [
        {
          "key": "Biodiversity",
          "doc_count": 2149642,
          "label": "Biodiversity",
          "is_selected": false
        },
        {
          "key": "Taxonomy",
          "doc_count": 1814489,
          "label": "Taxonomy",
          "is_selected": false
        }
      ],
      "label": "Subjects"
    },
    "file_type": {
      "buckets": [
        {
          "key": "pdf",
          "doc_count": 3633389,
          "label": "PDF",
          "is_selected": false
        },
        {
          "key": "png",
          "doc_count": 760182,
          "label": "PNG",
          "is_selected": false
        }
      ],
      "label": "File type"
    }
  },
  "links": {
    "self": "https://zenodo.org/api/records?page=1&size=1&sort=newest",
    "next": "https://zenodo.org/api/records?page=2&size=1&sort=newest"
  }
}