Interpreting SLP and mRNA exposure¶

The workflow answers a conditional question: given this starting material and these rates, how much of this target reaches APC surface MHC, where, and for how long? It supports free SLP, secretion-tagged mRNA-LNP and cytosolic-antigen mRNA-LNP after local injection. See core delivery and blood exposure for first-arrival fractions, endogenous expression and preformed pMHC transfer. It does not infer an effective immune dose from a peptide sequence.

Three quantities to read together¶

Question	Quantity	Interpretation
What remains available?	Target-bearing antigen copies over time	Includes shorter fragments retaining the chosen target; differs from intact-parent persistence
What is displayed now?	Surface pMHC copies in each APC compartment	Current complexes, after loading, trafficking and turnover; not copies per individual APC
How much display occurred?	Surface pMHC exposure in complex-hours over `[0,T]`	Integral of the curve; retain timing, location and horizon alongside the area

Use draining-node presentation as one clearly named endpoint, while inspecting local and systemic display too. Equal areas can have different peaks, timing and cellular distributions. The current model has no T-cell or antibody response equation, and no validated threshold defining an effective dose.

Run and inspect an explicit scenario¶

Start with the trajectory input contract. Select the exact target occurrence and allele, the delivery route and actual starting material. Supply measured rates only where the study and material apply; label sensitivity assumptions as assumed and intentional exclusions as disabled. A recognition score or binding rank is not an inverse-hour rate.

mhctools vaccine-trajectory --input scenario.json \
  --output trajectory.json --report trajectory.md

The report starts with the material, source normalization, context and horizon. It then shows presentation, antigen accounting and provenance. An incomplete scenario instead shows the missing steps and fragment assessments. Resolve those against evidence or declare explicit scenario assumptions; do not fill unknowns with zeros merely to obtain a curve.

The amount is material entering the model: released free peptide or carrier-bound RNA. Document copies, pmol or mass. For RNA mass, provide the full transcript's molecular weight. For an SLP, provide the actual peptide's molecular weight. Equal micrograms imply different molecule counts; equal molecule counts do not imply equal antigen production.

Compare matched endpoints¶

mhctools vaccine-compare --input slp-trajectory.json mrna-trajectory.json \
  --output comparison.json --report comparison.md

Inputs are result JSON files, not scenario input files. The comparison uses results from mhctools 3.50.0 or newer; rerun older scenario inputs to obtain integrated exposure and explicit dose accounting. The comparison requires matching target sequence, allele, MHC class and observation horizon. Original target coordinates can differ between constructs. Known organism differences are rejected; unknown context and route differences remain visible. Different MHC-II flanks or construct sequences can still affect processing and loading, so inspect each scenario's full fragment and rate audit.

The table preserves input order and shows absolute draining-node exposure next to exposure per input peptide/transcript. Its dose statement distinguishes equal documented starting mass, equal starting molecule counts, unequal amounts and one-input normalization. It does not sort by efficacy, invent a common probability or hide an unassessed scenario behind a zero. The comparison JSON retains rate values/sources and hashes of its input files; individual results retain all curves and cleavage assessments.

For sensitivity analysis, change a documented rate or starting amount in a copy of the scenario, rerun, and compare at the same horizon. Label the change in the scenario name and rate source. A spread across assumptions is not a confidence interval; no uncertainty distribution has been fitted.

Reproduce the complete demonstration¶

python scripts/vaccine_trajectory_example.py --output /tmp/vaccine-demo --plot
mhctools vaccine-trajectory --input /tmp/vaccine-demo/slp-input.json \
  --output /tmp/slp-check.json --report /tmp/slp-check.md
mhctools vaccine-compare \
  --input /tmp/vaccine-demo/slp-trajectory.json /tmp/vaccine-demo/mrna-trajectory.json \
  --output /tmp/comparison-check.json --report /tmp/comparison-check.md

The script writes input files, full results, individual readable reports, a comparison and optionally a plot (Matplotlib is included in the base install). Every rate is synthetic. Each panel labels its peptide/transcript denominator; the panels do not establish equal-dose efficacy. Use new output paths on each run.

What is grounded, and what still needs calibration?¶

Primary evidence	What it supports	What it does not identify
Moynihan et al., 2018	Formulation-dependent lymphatic distribution and presentation persistence in peptide-vaccine experiments	Universal free-SLP drainage or destruction rates
Liang et al., 2017	Separate LNP uptake and expression measurements in rhesus-macaque injection-site and draining-node APCs	Absolute target-specific pMHC kinetics for arbitrary human constructs
Dasti et al., 2025	An existing multiscale mRNA-vaccine model connecting modeled MHC-II exposure to cellular dynamics, with vaccine-specific calibration	Direct measurement of all latent presentation rates, or transferable SLP/secreted-mRNA defaults

The Dasti model uses macaque APC data and human antibody data. Its calibration is relevant prior work, but antibody agreement does not by itself validate every intermediate rate. This implementation does not import those fitted coefficients.

To calibrate the present model, pair time-resolved measurements of actual starting material, intact parent and target-bearing fragments, active intracellular RNA, completed/secreted antigen, and allele-specific surface pMHC in relevant APC locations. Retain cell counts when converting per-cell assays to compartment totals. Establish which rates are identifiable before fitting them, and evaluate independent constructs/donors/studies after checking source overlap. The serum audit supplies a bounded example of that evidence discipline; it does not calibrate lymph or APC kinetics.

The machine-readable vaccine_trajectory_evidence() inventory preserves these endpoint limits and calibration requirements. The calibration audit establishes the current non-identifiability and minimal matched measurements. Future biological calibration remains supported by independent benchmarks, uptake/presentation validation and node-exposure evidence. For primary mRNA mechanism studies and reported calibration observations, see the mRNA antigen-presentation bibliography.