| # | Layer | Failure if unguarded | Guard |
|---|---|---|---|
| 1 | REALITY Does modulating target T actually cause endpoint E? | On-target biology conflated with drug pharmacology. | unresolvable; declared in the model card |
| 2 | SELECTIONadded Which truths ever get the chance to become evidence? | Degree bias and confounding by indication are born here, not in the model. | audit/degree.py · data/attribution.py |
| 3 | EVIDENCE What have we actually observed? | Undated edges admitted, making temporal claims unverifiable. | data/ingest.py |
| 4 | ATTRIBUTIONadded Drug-level observation to target-level claim. Which target did it? | Uniform propagation makes the label a function of target degree. | data/attribution.py — noisy-OR + Mantel-Haenszel |
| 5 | LABELING RULE How do we interpret the evidence? | Unlabelled silently treated as negative; tiers pooled. | data/labels.py — three states, tiers never pooled |
| 6 | DATASET LABEL positive / earned-negative / unlabelled, per tier | Multiaxiality corrupts the label matrix. | data/ontology.py · models/hierarchy.py |
| 7 | SPLITadded Which labels does the model see, and which is it tested on? | An interpolation split behind an extrapolation claim. | data/splits.py · audit/leakage.py |
| 8 | MODEL What can the GNN learn from it? | A popularity detector with an excellent AUROC. | models/hetero_gnn.py — degree offset, partial rho |
| 9 | VERDICTadded Promote or reject? At what operating point, on what evidence? | A score shipped where a decision was required. | eval/metrics.py · pipeline.py |
The three nested questions
Each is stated in full rather than by label: the first draft of the PDF named Q2 and Q3 in a panel and defined them nowhere.
Q1 Is there any signal beyond node popularity?
Contrast: GNN vs a degree-only null: logistic regression on log-degree, no biology, no relations.
If the contrast ties: The knowledge graph encodes study attention, not safety biology.
Q2 Does the graph help beyond non-graph target features?
Contrast: GNN vs a tabular model on target annotations only: expression breadth, pLI and LOEUF, protein family.
If the contrast ties: Graph structure adds nothing. Ship the cheaper tabular model.
Q3 Do the relation TYPES carry the signal?
Contrast: Relation-aware (R-GCN, CompGCN, HGT) vs relation-agnostic (GraphSAGE, GAT on a collapsed adjacency).
If the contrast ties: Connectivity matters; relationships per se do not. This is the literal research question.