# Executive Summary — Personalized pMHC-II Tolerance Antigen-Selection Engine

**Paper:** *A patient-specific antigen-selection engine for personalized pMHC-II tolerance therapy in eosinophilic esophagitis.*
**Status:** bioRxiv-style preprint (Immunology; secondary Bioinformatics/Molecular Biology). **Preclinical-stage computational design study; all epitope activities are in-silico predictions, structural metrics are model-confidence estimates, and no tetramer or TCR validation has been performed. Not medical advice.**

## What the paper does
It reframes EoE's central obstacle — that patients react to multiple, patient-variable food allergens, so no single epitope can be fixed in advance — as a per-patient computation. It presents (1) an antigen-selection engine that converts a patient's HLA class-II genotype into a ranked, manufacturable epitope set; (2) AlphaFold-multimer structural validation of the selected pMHC-II complexes against explicit acceptance gates; (3) a shared-backbone, swappable-cassette construct design (the PACT™ concept) that makes per-patient manufacturing tractable by changing only the peptide; and (4) a tolerogenic multivalent nanoparticle scaffold for delivery.

## The core result
- Every approved and late-stage EoE therapy is a broad, chronically dosed suppressor; none is antigen-specific and none induces durable tolerance. Yet EoE is mechanistically antigen-specific — a food peptide presented on MHC-II to a pathogenic CD4⁺ T-cell clone.
- The engine turns HLA genotype into a ranked presentable-epitope set, and shows that antigen priority is genuinely individual and must be computed, not assumed.

## Key numbers (all traceable to artifacts/public sources)
- Scan: three dominant EoE allergens (β-casein, α/β-gliadin, β-conglycinin) × five common HLA-DRB1 alleles = 4,640 peptide × allele evaluations → **832 strong binders (IC₅₀ < 500 nM)**.
- Presented epitope load is strongly HLA-dependent: **13 to 352 strong binders per allele** across the same allergen panel — a 27-fold range that establishes antigen priority as per-patient.
- Worked example: a milk (β-casein) lead epitope on HLA-DRB1*07:01 clears an **ipTM ≥ 0.87** acceptance gate with full in-groove peptide engagement.
- Delivery: a 20 nm iron-oxide nanoparticle displaying five pMHC copies (~1.4% surface occupancy, ~24× avidity).

## Why it matters
It converts the "no single causal antigen" problem — the reason EoE has resisted a targeted tolerance therapy — into a tractable personalized-medicine workflow, coupled to a manufacturable, tolerogenic construct design. The framework is a platform template extensible to celiac and other food/auto-antigen disease.

## Honest verdict and what remains
The framing matches the evidence: the engine's output is a set of *predicted* binders, and MHC-II binding prediction does not by itself establish immunodominance, natural processing, or a tolerogenic (versus effector) outcome. No tetramer or TCR validation exists for any epitope; the milk lead is motivated by documented milk-responsive T cells in EoE, not a validated epitope–allele–TCR triad. Structural validation covers a single worked allele (DRB1*07:01) — extension to all five alleles and to more allergens (egg included) is prediction, not yet folded/validated design. The pMHC tolerance modality remains clinically unproven; the strongest support is mechanistic read-across from the Navacim and celiac (TAK-101) programs. Advancement is framed as a biomarker-first cascade (structural gate → biochemical stability → tetramer → IL-10-biased Tr1 signature → in-vivo/human PD), with a companion diagnostic as a regulatory precondition.