NNC2215 is glucose responsive.
Published receptor, cellular and animal studies establish glucose-responsive activity for NNC2215.
Public research status · updated 4 November 2026
Confidence before
candidate count.
We separate published facts, reproduced computation and untested chemistry. The program has a defined four-compound experiment, but no synthesized Dia Health One candidate and no candidate-specific binding result.
Evidence ladder
What is actually established?
Evidence levels run from E0 concept to E6 preclinical candidate. A claim cannot inherit confidence from another molecule or another observable.
PaG makes NNC2215 non-viable.
Direct binding, receptor assays, pig-clamp data and modeling establish a clinically relevant PaG liability.
MC-Q1 identifies where to test.
The contact-versus-distal distinction survives denser geometric sampling. This does not predict affinity.
MC-Q1 glucose binding.
No material exists. Public methoxy-roof data are a negative prior, not confirmation.
MC-Q1 PaG selectivity.
No direct binding or calibrated prospective free-energy result exists.
Working insulin candidate.
No MC-Q1 reporter switch or insulin conjugate has been synthesized or functionally tested.
Claim ledger
Current position.
“Unknown” is a program state. Each next gate is designed to replace one important unknown with a calibrated or measured result.
| Claim | Status | Current evidence | Result needed |
|---|---|---|---|
| NNC2215 glucose response | HIGH CONFIDENCE | Published in-vitro and in-vivo data | Used as benchmark, not as our candidate |
| PaG liability | HIGH CONFIDENCE | PaG binds NNC2215 more strongly than glucose and affects activity at clinically relevant exposure | New design must reject PaG and broader glucuronides |
| MC-Q1 local mechanism | MODERATE / MODEL-DEPENDENT | Rigid geometry, 13,824-pose stress test and positional control | Calibrated host–guest free energy and direct binding |
| MC-Q1 synthesis | NOT CONFIRMED | Exact graph and provider-ready specification | Accepted route, regioisomer assignment, quote and material specification |
| MC-Q1 glucose retention | UNKNOWN | No measurement; methoxy carries a negative public prior | Same-run glucose binding versus MC-P0 |
| MC-Q1 PaG rejection | UNKNOWN | No direct result | Kd(PaG)/Kd(glucose) ≥ 5, lower 95% bound >1 |
| Reversible switch | NOT BUILT | Specification only | Non-insulin reporter, down-step reversibility, HSA and interferent panel |
| Functional insulin | NOT BUILT | No conjugate | Human-cell signaling, potency, Emax, selectivity and PaG resistance |
Minimum information package
Four molecules, one causal test.
The goal is not a large library. It is to distinguish contact-site chemistry from generic polarity and position-independent effects.
Parent
MC-P0Same-assay reference for glucose and PaG binding.Primary hypothesis
MC-Q1Contact-site methoxy. Local polarity/electronic/solvation hypothesis—not a steric-block claim.Positional control
MC-Q1DThe same methoxy edit at distal roof site A.Constitutional control
MC-P3Contact-site hydroxymethyl; separates topology from generic polarity.Compute-plan correction
Do not buy false precision.
Boltz-2 is not the affinity engine for this problem. Its documented affinity mode is for a small-molecule ligand against a protein target. GluHUT binding glucose or PaG is a synthetic host–guest system.
Boltz also does not accept concentration as a physical variable. Repeating predictions with labels such as 3, 5, 10 and 20 mM cannot create a Hill curve.
The corrected AWS plan first reproduces the public GluHUT explicit-solvent umbrella-sampling benchmark, then calibrates parent glucose-versus-PaG behavior, and only then compares MC-Q1 and controls prospectively.
Confidence-changing work
The next gates.
Every gate has a stop rule. More compute is not progress when the method or chemical object has not passed the preceding gate.
C1 · Reproduce public GluHUT analysis
Reanalyse deposited trajectories and regenerate published PMFs and uncertainty before generating new trajectories.
C2 · Validate chemistry and force field
Resolve exact parent graph, charges, nonstandard terms, protonation and PaG treatment.
C3 · Calibrate parent glucose versus PaG
Use multiple explicit-solvent paths. Require stable direction, overlap and uncertainty smaller than the effect.
C4 · Prospective candidate comparison
Without retuning, compare MC-P0, MC-Q1, MC-Q1D, MC-P3 and preferably MC-P1.
C5 · Provider acceptance
Obtain a written route, positional assignment, material specification, price and rights/data terms.
C6 · Direct binding
Use two orthogonal methods. This is the first gate capable of turning MC-Q1 into a measured hit.
Open records
Inspect the evidence.
The records expose decision logic, limitations and stop rules. Negative results remain part of the archive.