Encounter Check v0.2.0-alpha

Try the model: https://datarichinsightpoor.github.io/encounter-check/
The math: https://datarichinsightpoor.github.io/encounter-check/methods.html and web/MATH.md
Model code: web/model.js

All presets and defaults are synthetic.

No. 05 in 99 Small Problems. First public research-use alpha. Version 0.1.0-alpha was a private preview and was never published; its notes are kept below for the development record.

In this release

A bispecific drug now links effector to target, so dose and density interact.

• Optional drug step (04 in the form). Monovalent equilibrium binding to antigen and effector-receptor pools, free drug solved from the mass balance by bisection, and a normalized bridge index that peaks at free drug (\sqrt{K_TK_E}). The kill probability per conjugate becomes (p_{max}(1-e^{-sB})). Contact equations are unchanged.
• New section, Discriminating experiment 3: dose response at three densities with and without drug bound by cells, plateau, free and total EC50, drug on cells at EC50, free fraction and lysis at the declared dose, a contact-versus-depletion decomposition, and a volume probe.
• Drug indices in the indices table; the contact rival reports its kill probability at the declared dose.
• Plate map gains a dose axis and free-fraction column when the drug is on.
• New exports: dose figure PNG and dose-response CSV.
• New synthetic preset: "Same ratio, same dose."
• Math contract gains a bridge section with two new references (Douglass et al. 2013; Schropp et al. 2019).

Synthetic default results (bispecific preset)

At 1:1 in a 32 mm², 100 µL settled well over 4 h, across 0.5×, 1× and 2× density: plateau 27.7%, 40.5% and 53.5%. Free EC50 1.72, 1.60 and 1.47 pM. Total EC50 2.94, 3.89 and 5.74 pM. At 5 pM total, free fraction 61%, 42% and 25%, and lysis 18.4%, 23.5% and 24.5%. Halving the medium at fixed cells and concentration lowers lysis from 23.5% to 17.6% with binding and leaves it at 32.5% without.

Verification

58 of 58 tests pass. Browser QA at 1360 and 375 px in dark and light themes; no console errors. See VERIFICATION.md.

Boundaries

No confidence coverage, power or cost weighting. Rate constants and affinities are declared, not fitted. Software verification is not biological validation.