# Personalized Therapy Design — Index-Case EoE Patient *Eosinophilic esophagitis, computational hypothesis-generating design. Built from the patient's HLA class-II typing and a 7-SNP allergy/eosinophil genetic screen, on the Phase 1–4 EoE target-discovery and pMHC-therapy platform.* --- ## ⚠️ Scope and disclaimer This is a **research / hackathon design exercise**, not medical advice and not a clinical treatment plan. Every epitope here is a **computational prediction** (mhcnuggets class-II binding affinity); every target rationale is **hypothesis-generating**. None of it has been validated in this patient's cells or in any wet-lab assay. Nothing here should guide this individual's care. Real decisions require a qualified clinician with the full patient record, and any therapeutic step requires prospective HLA-matched T-cell validation and formal preclinical/clinical development. HLA typing, genotypes, and phenotype were provided by the user for this exercise. --- ## 1. Patient molecular + clinical profile **Clinical phenotype:** full atopic march — eczema, allergic rhinitis, asthma, environmental + food allergy, and eosinophilic esophagitis. This multi-organ type-2 picture is central to target selection below. **HLA class II (2 haplotypes → 4 functional MHC-II molecules):** | Haplotype | DR molecule | DQ molecule | |---|---|---| | 1 (DR7-DQ2.2) | DRB1\*07:01 | DQ2.2 = DQA1\*02:01 / DQB1\*02:02 | | 2 (DR15-DQ6.2) | DRB1\*15:01 | DQ6.2 = DQA1\*01:02 / DQB1\*06:02 | **DQ2.2 is the celiac-disease-associated gluten-presenting heterodimer** (Sollid gluten-epitope nomenclature; Fallang 2009; Bergseng 2014). This directly shapes how this patient handles wheat — see §2. **Genetic screen (7 SNPs):** heavy eosinophil-elevating load. Four of five eosinophil-axis variants are risk-elevating — including *GATA2* (rs9880192) and *CEBPE* (rs2239633), two master transcription factors of eosinophil differentiation — with only *IKZF2* (rs12619285) protective. Both allergic-axis variants are risk: *RERE* (rs301806), whose GWAS-Catalog trait list literally includes eczematoid dermatitis / allergic rhinitis / allergic disease, matching this patient's comorbidities, and a shared-atopy locus near *GSAP* (rs4296977). Full annotation in `patient_gwas_annotation.csv`. --- ## 2. Which allergens/epitopes to target in THIS patient Personalized epitope panel: the 10 food-allergen proteins were scored against **all four** of this patient's MHC-II molecules (2,424 unique 15-mers × 4 molecules = 9,696 predictions; `patient_epitope_panel.csv`). The patient presents **1,087 strong-binding epitopes** (IC50 < 500 nM) in total. **Presentation load by molecule:** DR7 356, DR15 324, DQ6.2 279, DQ2.2 128. **Top-presented allergens (summed across the 4 molecules):** | Allergen | Total strong binders | Notable | |---|---|---| | Soy β-conglycinin | 278 | highest overall; DR7/DR15-driven | | Soy glycinin | 248 | DR7/DR15-driven | | **Wheat α-gliadin** | **204** | **88 on DQ2.2 alone — the celiac signature** | | Milk β-lactoglobulin | 115 | broad | | Milk α-S1-casein | 66 | DR15-driven | **The DQ2.2-gliadin finding is the single most patient-specific result.** Wheat gliadin is presented far more heavily by DQ2.2 (88 strong binders) than by any of the patient's other three molecules (28–51). The top DQ2.2/gliadin cores are glutamine-rich (e.g. `LHQQQQQQQQQQQQP`, `QQQQQILQQILQQQL`) — the characteristic gluten-epitope register. This patient's genetics predispose them to present gluten peptides on the classic celiac heterodimer, so **wheat is a mechanistically prioritized antigen for this individual** in a way it would not be for a DQ2.2-negative patient. (This does not diagnose celiac disease — it means the antigen-presentation machinery for gluten is present and should be factored into both dietary counseling and any antigen-directed reagent.) **Antigen priority for this patient:** soy (β-conglycinin, glycinin) and wheat gliadin lead, followed by milk. Shellfish/peanut/tree-nut are presented weakly on this patient's alleles. --- ## 3. Personalized pMHC-anergy reagent design The platform builds **patient-matched pMHC-II reagents** — the dominant predicted core loaded onto the patient's own MHC-II molecule — to engage and tolerize (anergize) the cognate allergen-specific CD4+ T cells (mechanism: `pmhc_tcr_mechanism_review.md`, Mechanism 1). For this patient the lead reagents are: | Molecule | Lead allergen core | Predicted IC50 (nM) | |---|---|---| | DR7 (DRB1\*07:01) | Soy β-conglycinin `ALLLPHFNSKAIVIL` | 2.1 | | DR15 (DRB1\*15:01) | Peanut Ara h 2 `LTILVALALFLLAAH` | 6.7 | | DQ2.2 (DQA1\*02:01/DQB1\*02:02) | Soy β-conglycinin `RYDDFFLSSTQAQQS` | 34.0 | | DQ6.2 (DQA1\*01:02/DQB1\*06:02) | Peanut Ara h 2 `LTILVALALFLLAAH` | 23.6 | Plus a **DQ2.2/gliadin reagent** (top gluten core, e.g. `IILHQQQQQQQQQQQ`) to address the patient's dominant wheat-presentation axis. Because the patient is compound-heterozygous, a complete reagent set spans all four molecules; the two DR molecules carry the highest predicted load and are the primary anergy targets, with the DQ reagents (especially DQ2.2-gliadin) adding patient-specific coverage. --- ## 4. Effector / receptor-target arm — which of IL-33/ST2, CCL26, SIGLEC6 fits this patient Antigen-directed tolerization addresses the *trigger*; the three Phase-3 protein-design targets address the *effector inflammation*. Scored against this patient (`patient_target_assessment.csv`): **IL-33 / ST2 (IL1RL1) — fit 5/5, SYSTEMIC lead.** IL-33 is the epithelial alarmin at the apex of the type-2 cascade, driving asthma, atopic dermatitis, rhinitis, and EoE (Saikumar Jayalatha 2021; Chan 2019; Cayrol 2017). For a patient with the **whole atopic march**, this is the one target that reaches every organ involved, not just the esophagus. Receptor-side IL1RL1 blockade (the Phase-3 lead, 1,781 Ų validated interface, isoform-independent) is the strategically broadest arm. > *Honest caveat:* the patient's 6p21 variant rs1131896 is in the **HLA-B/MICA** region, **not** *IL1RL1* (which is at 2q12). It is not a genetic tag for the ST2 target. The IL-33/ST2 rationale rests on the patient's multi-organ atopic phenotype and the pathway biology, not on these seven SNPs. **CCL26 / eotaxin-3 — fit 5/5, best genotype match.** CCL26 is the dominant eosinophil-recruiting chemokine in EoE tissue (Blanchard 2006). This patient carries four eosinophil-**elevating** loci (*SH2B3*, the HLA region, and the eosinophil master-TFs *GATA2* and *CEBPE*) — genetics that **supply** eosinophils, which CCL26 then **traffics** into esophageal tissue. Blocking CCL26 (or the CCR3 axis) is the effector step most tightly matched to this individual's genetic eosinophil load. **SIGLEC6 — fit 3/5, EoE-local arm.** SIGLEC6 marks the EoE-specific disease-associated mast-cell state (Phase-2: 84.5% vs 34.2%; Robida 2022). Tissue-restricted — a valuable depleting/ADC modality for the esophageal mast compartment, but it does not address the systemic march. Complements rather than leads. --- ## 5. Integrated combination rationale For this patient the design is **two-armed**: 1. **Upstream, antigen-directed (personalized pMHC anergy):** tolerize the soy-, wheat-gliadin-, and milk-specific CD4+ T cells presented on the patient's own DR7/DR15/DQ2.2/DQ6.2 molecules — removing the antigenic drive. 2. **Downstream, effector blockade (patient-selected target):** **IL-33/ST2** as the systemic lead (covers the full atopic march), with **CCL26** as the EoE-effector arm matched to the patient's eosinophil-elevating genotype, and **SIGLEC6** reserved for esophagus-local mast-cell control. The genetic screen tips the balance toward pairing antigen-directed therapy **with** effector blockade rather than antigen-directed therapy alone: this patient's eosinophil program is genetically reinforced (GATA2/CEBPE/SH2B3), so removing the trigger may be necessary but not sufficient, and a downstream effector target (CCL26 and/or IL-33/ST2) is expected to be additive. --- ## 6. Ex vivo per-patient validation path Before any therapeutic inference, the design is testable on this patient's own cells using the Task 3 assay (`t_cell_assay_protocol.md`), HLA-agnostic and per-patient by construction: 1. Isolate PBMC / esophageal-biopsy T cells. 2. Expand against **this patient's** predicted dominant cores (soy β-conglycinin, wheat gliadin, milk) — measure proliferation (SI), Th2 cytokines (IFNγ/IL-5/IL-13), activation (CD25/HLA-DR). 3. Confirm the predicted immunodominance hierarchy (does measured response track the predicted per-molecule burden? DQ2.2-gliadin reactivity present?). 4. Build patient-matched pMHC-II multimers for the lead cores; test tetramer engagement and anergy induction ex vivo. 5. TCR-sequence dominant allergen-specific clones. Predicted epitope burden is a **mechanistic covariate to be validated**, never an eligibility gate — the platform is designed so that any patient, of any HLA type, gets a personalized panel computed from their own genotype. --- ## 7. Limitations - **Epitopes are computational** (mhcnuggets class-II affinity); no experimental confirmation of presentation or immunodominance in this patient. Cross-validated predictor concordance was ρ≈0.79 on shared DR alleles, but DQ predictions are less benchmarked than DR. - **The 7-SNP screen is a small, curated panel**, not a genome-wide PRS; the "genetic load" read is qualitative. Gene assignments are nearest-gene (Ensembl/GWAS-Catalog), not fine-mapped causal genes. - **rs1131896 is HLA-region (HLA-B/MICA), not IL1RL1** — no direct genetic support for the ST2 target in this patient (see §4). - **No wet-lab, no clinical data.** The two-armed design is a hypothesis; the ex vivo path in §6 is the first real test. - Target-fit scores are structured expert judgment over retrieved literature + prior-phase evidence, not a quantitative model output. --- ## Key references (DOI-verified, retrieved via OpenAlex/GWAS-Catalog this session) - Sollid LM et al. 2012, *Immunogenetics* — gluten T-cell epitope nomenclature (DQ2.2 relevance). doi:10.1007/s00251-012-0599-z - Fallang L-E et al. 2009, *Nat Immunol* — DQ2.5 vs DQ2.2 celiac risk. doi:10.1038/ni.1780 - Bergseng E et al. 2014, *Immunogenetics* — DQ2.2 binding motif. doi:10.1007/s00251-014-0819-9 - Kottyan LC et al. 2017, *Mucosal Immunol* — genetics of EoE. doi:10.1038/mi.2017.4 - Ferreira MAR et al. 2017, *Nat Genet* — shared genetic origin of asthma/hay fever/eczema. doi:10.1038/ng.3985 - Saikumar Jayalatha AK et al. 2021, *Pharmacol Ther* — IL-33/IL-1RL1 in asthma. doi:10.1016/j.pharmthera.2021.107847 - Chan BC-L et al. 2019, *Front Immunol* — IL-33 in allergic inflammation. doi:10.3389/fimmu.2019.00364 - Cayrol C & Girard J-P 2017, *Immunol Rev* — IL-33 alarmin biology. doi:10.1111/imr.12619 - Blanchard C et al. 2006, *J Clin Invest* — eotaxin-3/CCL26 in EoE. doi:10.1172/jci26679 - Robida PA et al. 2022, *Cells* — Siglec-6 on human mast cells. doi:10.3390/cells11071138 - (Per-SNP citations in `patient_gwas_annotation.csv`.)