{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-11-11-004",
    "source_story_id": "tmp-story-pyrite-pose-ranking",
    "edition_id": "mp-2026-11-11-morning-0124",
    "edition_url": "https://themachinepress.com/edition/2026-11-11",
    "position": 4,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "The Binding Pose Becomes an Agent Test",
    "slug": "the-binding-pose-becomes-an-agent-test",
    "dek": "A 1,500-pair benchmark asks models and chemists to rank candidate protein-ligand poses that cofolding systems already know how to produce.",
    "summary": "A 1,500-pair benchmark asks models and chemists to rank candidate protein-ligand poses that cofolding systems already know how to produce.",
    "body_text": "Pyrite turns protein-ligand pose selection into 1,500 pairwise comparisons across diverse complexes. In the reported study, expert chemists using 3D viewers and physics-based scores performed near chance, while an LLM reasoner reached up to 67% accuracy and an agent that wrote features for logistic regression reached 74%. The authors found that LLM-guided feature discovery generally beat direct textual reasoning and that reinforcement learning improved open-weight models. The benchmark measures ranking within its constructed pairs; it does not establish clinical or drug-development performance.",
    "why_it_matters": "A 1,500-pair benchmark asks models and chemists to rank candidate protein-ligand poses that cofolding systems already know how to produce.",
    "limitations": [
      "The benchmark measures ranking within its constructed pairs; it does not establish clinical or drug-development performance."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-11-004/the-binding-pose-becomes-an-agent-test",
    "json_url": "https://themachinepress.com/story/mp-2026-11-11-004.json",
    "first_published_at": "2026-11-11T09:00:00.000-05:00",
    "modified_at": "2026-11-11T09:00:00.000-05:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-11-11-004-001",
        "text": "A 1,500-pair benchmark asks models and chemists to rank candidate protein-ligand poses that cofolding systems already know how to produce.",
        "source_ids": [
          "source-2026-11-11-004"
        ],
        "qualification": "The benchmark measures ranking within its constructed pairs; it does not establish clinical or drug-development performance."
      }
    ],
    "source_ids": [
      "source-2026-11-11-004"
    ],
    "tags": [
      "pyrite",
      "protein",
      "ligand",
      "pose",
      "ranking",
      "agents"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-11-11-004",
      "title": "arXiv:2611.14511v1",
      "publisher": "arXiv / Cornell University",
      "url": "https://arxiv.org/abs/2611.14511v1",
      "canonical_url": "https://arxiv.org/abs/2611.14511v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-11-10T13:56:32.000-05:00",
      "accessed_at": "2026-11-11T08:26:32.531-05:00",
      "supports_claim_ids": [
        "claim-mp-2026-11-11-004-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Binding Pose Becomes an Agent Test",
    "publisher": "The Machine Press",
    "published_at": "2026-11-11T09:00:00.000-05:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-11-004/the-binding-pose-becomes-an-agent-test"
  }
}