Loading evidence layers…
Resolving job, evidence matrix, model agreement, and quantum circuit.
Resolving job, evidence matrix, model agreement, and quantum circuit.
Five focused views — a 2D molecular sketch, the Pauli feature-map quantum circuit, the compound → gene → disease evidence path, the six-layer evidence matrix, and per-family model agreement. All rendered as static SVG so the lite build stays small. The full version replaces these with interactive 3D / live-job-driven counterparts.
Stylised 2D structural sketch — fused aromatic system + amide side chain + CF₃ group + hydroxyl handle. The full version renders an interactive 3D conformation via WebGL; this lite view is a static SVG that ships zero extra JS.
The headline encoding circuit. Hadamards prepare uniform superposition, single-qubit RZ rotations encode one data feature each, then pairwise CNOTs entangle adjacent qubits and a second RZ layer encodes pair-wise feature products. Two repetitions deepen the Hilbert-space embedding before measurement.
A walk through Hetionet — compound, target, pathway, disease — with weights from the trained model. The path must aggregate to at least the threshold (0.55) to count as mechanistic support.
Each row is an evidence layer; each cell is a finding from that layer with a state — live (real source returned data), supports / weakens (the score moves the right way), fallback (synthetic stand-in), or missing (no signal at all).
Top score per family vs the cross-family mean. Spread is the signal: small spread means classical/hybrid/quantum agree, large spread means you should probably not bet the run on a single family.