First, check what was actually drawn
The C4 branch began with a peroxide-bearing moxidectin conjugate. Its description called the added ring a trioxane. Checking the deposited structure showed a different ring: a 1,2-dioxane. The linker description also needed correction. Those distinctions matter because the structure determines which stability and release arguments apply.
The correction was published with the redesign. The resulting paper compares 34 proposed successors and records the costs of the alternatives: molecular size, calculated lipophilicity, polarity, stereochemical ambiguity and the difficulty of preserving the peroxide during synthesis. None of those calculated descriptors establishes biological activity.
More accumulation is not the same as more selectivity
The B1 branch examined a different idea: attach a positively charged targeting group to influence mitochondrial distribution. Its review changed four conclusions. A more aggressive monovalent cation does not change the ideal Nernst slope. Depolarisation can reduce further accumulation. A weak base can be trapped in acidic compartments instead of reaching the intended mitochondrial destination. And changing an arylboronate into a phenol does not remove a phosphonium charge.
These are design decisions, not small editorial changes. Each correction changes what should be built, compared or rejected. The published paper explains the calculations and reports the cross-check between the two research nodes.
A model you can turn is a model you can question
The stage above displays the documented moxidectin parent scaffold. Its coordinates were generated with RDKit distance geometry and local force-field optimisation. The coloured spheres and bonds show one computed three-dimensional conformer. The tetrahedral cage is a display guide; it is not a claim that the molecule itself is a tetrahedron.
Drag to inspect another angle, switch to wireframe, or download the structure. A computed conformer gives a concrete object to discuss. It is not an experimental crystal structure, a simulation of binding, or proof of how the molecule behaves in a person.
The deliverable is the decision trail
The value of the project is the connected record: a defined structure, a stated rationale, computed properties, independent checks, rejected alternatives and explicit tests that could overturn the design. The structures remain proposals. Synthesis, stability, distribution and activity still require experimental measurement.
This is also the principle behind the Research Factory workspace below. A mission stays on the left; reports, evidence, drawings and the ledger stay on the right. The demonstration shows the intended review process. Saving a brief does not yet activate paid research or assign the council.
Read the source record
- C4 successor paper: structural correction, 34 designs and decision gates (PDF)
- B1 successor paper: targeting arithmetic, alternatives and four corrected conclusions (PDF)
- Moxidectin parent: the stage’s computed 3D structure (SDF)
Editorial source: the public branch papers. The original issue, research articles and publication archive are preserved below. This report describes computational designs, not a demonstrated cancer treatment.
