The plate floor is coated with capture antibodies. Biomarker antigens (orange) diffuse through the well by Brownian motion; when one drifts close to a free capture site, it binds with a probability set by affinity and sticks in place. A second wave of enzyme-linked detection antibodies (cyan) then diffuses in and can only bind an already-captured antigen — completing the "sandwich". Only fully sandwiched sites turn the signal color, so the readout reports real biomarker, not just anything that happened to land nearby.
bound(C) = C / (C + Kd) Langmuir isotherm
P(bind per Δt) = affinity · Δt (per nearby collision)
signal (OD) = sandwiched_sites / total_sites
result = POSITIVE if signal ≥ threshold else NEGATIVE
- Biomarker concentration — how many antigen molecules diffuse into the well; higher concentration means more collisions with capture sites per second.
- Antibody affinity — the binding probability per close encounter, for both the capture and detection step; low-affinity antibodies need far more collisions (or higher concentration) to light up a site.
- Threshold — the OD cutoff a real assay is calibrated against; push it up and you demand a stronger signal before calling a sample positive, trading sensitivity for specificity.
- Signal (OD) — the fraction of capture sites carrying a complete antigen + detection-antibody sandwich, standing in for the optical density a real plate reader measures.
Real-world relevance: this two-step "capture, then detect" logic is exactly why sandwich ELISA (used for PSA, troponin, HER2 and hundreds of other clinical biomarkers) is far more specific than a single-antibody assay — a false hit has to survive two independent binding events, not one.