SESTRAV: Structural Epitope Scoring via TCR Recognition And Vaccinology

SESTRAV is a dry-lab (purely computational) pipeline for prioritizing viral CD8+ T-cell epitopes by predicted immunogenicity, covering nine viral pathogens (CMV, EBV, HBV, HCV, HPV, HIV-1, IAV, DENV, SARS-CoV-2). It targets the specificity bottleneck that binding-only tools leave open: MHC binding is a weak proxy for T-cell immunogenicity, so SESTRAV scores peptides on the physicochemical structure of TCR-contact residues (positions p4-p8) in addition to multi-allele presentation.

The system is organized as two model tracks under a single reproducible Snakemake workflow:

• Production track (validated): a Random Forest / XGBoost ensemble over a 31-feature structural representation (canonical mode_31). This is the maintained, benchmarked scorer.
• Research track (gated): a graph neural network (GINEConv + ESM-2 residue embeddings) that fuses a peptide residue graph with the same physicochemical features. It is the v2.0 forward architecture and is held to explicit promotion gates before it can become canonical.

SESTRAV carries the OpenSSF Best Practices Passing badge (project 13191) with a documented roadmap toward the Silver and Gold tiers (see Security and Compliance Posture). All results reported here are computational; no wet-lab efficacy is claimed. The end-to-end design is documented in ARCHITECTURE.md.

SESTRAV vs Field

Capability	SESTRAV	PredIG	PRIME	NetMHCpan	pVACtools
End-to-end workflow (proteome → ranked output)	✓	Partial	Partial	✗	✓ (neoantigens)
Open source, pip-installable	✓	✓	✓	Academic license	✓
Cryptographic dataset governance (freeze mode)	✓	✗	✗	✗	✗
OpenSSF Passing badge	✓	✗	✗	✗	✗
Antigen processing as training features	feature_mode=33	✓	Partial	✗	Partial
Graph Neural Network scorer	✓ (v2.3 GINEConv+ESM-2; research/ensemble component)	✗	✗	✗	✗
Pan-allele training	v5 active	Partial	✓	✓	✓
Multi-virus support	9 viruses (v5 active): CMV, EBV, HBV, HCV, HPV, HIV-1, IAV, DENV, SARS-CoV-2	Limited	Limited	Pan-pathogen	Tumor
Wet-lab candidate protocol included	✓	✗	✗	✗	Partial
AUC-PR on labeled benchmark (Tier A)	0.828 (OOF, full_31/mode_31)	not benchmarked	not benchmarked	N/A	N/A

Tier A 704-peptide labeled benchmark. SESTRAV RF is evaluated strictly out-of-fold (conservative); external tools are fully scored on the same peptides. The certified head-to-head field is in External Benchmark Results below - BigMHC (0.822), MHCflurry binding-only (0.800), MixMHCpred 2.2 (0.795), and DeepImmuno (0.698), all bound to results/table3_tier_a_metrics.csv; the closest external tool is BigMHC (0.822). PredIG and PRIME are compared on capabilities only: their metric head-to-head is not reproducible from a certified results file and is not reported. Separately, on the harder v5 generalization set (35,597 active rows, 9 viruses), canonical mode_31 scores self-proteome Gate 1 AUC-PR 0.8897 and reports same-pathogen (within-virus) discrimination per-virus (mean within-CV AUC-ROC 0.751; the previously reported pooled AUC-ROC 0.9368 was decoy-inflated and is retracted - see Paradigm 2 below). full_31/mode_31 (0.828) is the canonical track and the certified Tier-A head-to-head result; the extended full_33 antigen-processing configuration is reported separately under Release Tracks and is not part of this certified field. Methodology: docs/external_testing/External_Validation_Sign_Off.md.

---

Predicting whether a viral peptide will elicit a CD8⁺ T-cell response is harder than predicting MHC binding. Binding-only approaches achieve AUC-PR ≈ 0.80 on the SESTRAV benchmark - yet most public tools stop there. SESTRAV bridges this gap by extracting physicochemical features from TCR-contact residues (positions p4-p8, following Chowell et al. 2015) and training ensemble classifiers on experimentally validated IEDB immunogenicity data.

SESTRAV is a governed computational workflow for viral T-cell epitope prioritization (immunogenicity scoring against a self-proteome background). It integrates six computational stages - proteome-scale peptide generation, multi-allele MHC binding prediction, TCR contact physicochemical feature extraction, antigen processing scoring, ensemble immunogenicity inference, and freeze-mode governed output - under a single reproducible Snakemake DAG with cryptographic dataset provenance. To our knowledge, no publicly available tool integrates antigen processing, physicochemical TCR features, and graph neural network scoring within an OpenSSF-compliant, auditable pipeline.

The canonical release uses a 31-feature model (20 physicochemical properties at TCR-contact positions + 10 per-allele MHC binding scores + peptide length as the critical mediating variable). On the Tier A labeled benchmark it achieves AUC-PR 0.828 (weighted OOF; 0.864 unweighted ablation); on the harder v5 generalization set (35,597 active rows, 9 viruses) it achieves AUC-PR 0.8897 self-proteome Gate 1 and reports same-pathogen (within-virus) discrimination per-virus (mean within-CV AUC-ROC 0.751; the previously reported pooled AUC-ROC 0.9368 was decoy-inflated and is retracted - see External Benchmark Results). Leave-one-virus-out (LOO) cross-virus analysis (Amendment 7 corrected, IEDB assay-confirmed negatives only) yields mean AUC-ROC 0.463 with 3 of 9 viruses above chance - the model is designed for within-virus epitope prioritization and self-proteome discrimination, not cross-virus transfer; LOO results are reported in full below. Optional tiers add antigen processing features (NetChop/TAPreg, feature_mode=33) and a GINEConv+ESM-2 graph neural network research track.

Background and Motivation

Most computational pipelines focus on MHC presentation, predicting whether a peptide is displayed on the cell surface. However, binding affinity alone is a weak proxy for immunogenicity (typical AUC ≈ 0.60 when used directly; Carri et al., 2023). SESTRAV addresses this limitation by extracting features from TCR-contact residues (primarily positions p4-p8) and training classifiers on experimentally validated immunogenicity data from the IEDB.

This approach combines structural insights with multi-allele binding predictions to better discriminate true immunogenic epitopes.

Release Tracks and Policy

• Canonical track (default): 31-feature configuration (20 physicochemical + 10 multi-allele MHC binding + peptide length). Tier A AUC-PR 0.828 (weighted OOF) / v5 self-proteome Gate 1 AUC-PR 0.8897. Within-virus (same-pathogen) discrimination is reported per-virus (mean within-CV AUC-ROC 0.751; see Paradigm 2); the previously reported pooled AUC-ROC 0.9368 was decoy-inflated and is retracted. This is the maintained release path and the production scorer.
• Extended track: 33-feature configuration adds NetChop 3.1 and TAPreg antigen processing scores as training features (feature_mode=33). AUC-PR 0.886 (unweighted) / 0.840 (weighted) - v3 extended-track result (antigen-processing features); not part of the certified v5 Tier-A head-to-head. Requires antigen processing cache; see scripts/precompute_antigen_processing.py.
• Legacy comparator track: 30-feature (without peptide length) and 21-feature (sequence-only) configurations retained for historical reproducibility.

Source of Truth: SESTRAV v2 designates this repository (main branch) as the single authoritative source. For release-grade reproducibility, enable freeze_mode: true in config.yaml. Freeze mode enforces strict guardrails: no Stage 4 prototype fallback, no mixed legacy/canonical output stems, and atomic artifact updates.

Security & Compliance Posture

SESTRAV 2.0 maintains a rigorous security posture suitable for biomedical data pipelines.

• SAST & CI: All commits are gated by Bandit, CodeQL, and Semgrep via GitHub Actions.
• Dependency Pinning: Environment files use strict --require-hashes to mitigate supply-chain attacks.
• Data Integrity: The pipeline uses freeze_mode constraints to guarantee data immutability during reproducibility benchmarking.

OpenSSF Best Practices

Tier	Status	Evidence and remaining gap
Passing	Attained (project 13191)	Full criteria-to-evidence mapping in docs/openssf_best_practices_readiness.md.
Silver	Substantially met	Documented governance, two-scope CI coverage gating, Sigstore-signed releases, and a published threat model are in place. Remaining gap: the multi-person criteria (bus_factor, two_person_review, contributors_unassociated), which require a second maintainer.
Gold	Coverage targets met; in progress	Library-scope statement and branch coverage already clear the Gold thresholds (>= 90% statement, >= 80% branch; currently ~99% / ~98%). Remaining gaps: the multi-person criteria above, plus per-file SPDX/copyright headers (license_per_file), deferred until a second contributor lands so authorship is attributed accurately.

Tier progress is tracked in ROADMAP.md; governance and assurance evidence is in docs/threat_model.md and GOVERNANCE.md.

Coverage is gated at two scopes, deliberately, and the two numbers are not the same: unit statement/branch coverage on the importable library surface (currently ~99% / ~98%, clearing the Gold thresholds of >= 90% / >= 80%), and whole-repository statement coverage including the pipeline/CLI research scripts (gated at a lower floor of ~35% in pyproject.toml). The gap is by design - executable research scripts are validated by the integration and data/benchmark CI gates rather than by unit statement coverage (see .coveragerc.library).

For vulnerability reporting, refer to SECURITY.md. For a detailed compliance matrix against OpenSSF standards, see docs/security_compliance.md.

Validation Status

The committed release evidence (v3 dataset, 1004 peptides, 3.35:1 class ratio) provides the following computational validation:

Metric / Check	Result
H2 Tier A decision (R10 >= 2.0)	Not supported (R10 = 0.9494)
Gold-standard positive recovery	15/15 positives found; 7/15 in top 25%
Binding-only baseline comparison	Baseline recovers 15/15 (expected for strong-binder set)
Gold-standard negative discrimination	9/10 negatives pushed down vs. binding-only
SHAP feature contribution	~60% MHC binding, ~40% TCR-contact features

Important: These results constitute computational validation only and do not establish biological efficacy. Wet-lab experimental confirmation is required for any therapeutic claims. See results/final_validation_report.md and docs/limitations_statement_v1.md for full details.

External Benchmark Results

SESTRAV is evaluated under two complementary paradigms: (1) a Tier A labeled benchmark for a clean head-to-head against the field, and (2) a larger, harder hard-decoy generalization set that decouples MHC binding from immunogenicity. The two numbers are not competing - the v4 figure is lower by design because the task is harder.

Paradigm 1 - Tier A head-to-head (N=720 labeled; SESTRAV OOF on the N=704 scored intersection)

Tool	AUC-PR	ISSR@10	Evaluation
SESTRAV RF (full_31/mode_31, canonical)	0.828	0.843	OOF 5-fold (conservative)
BigMHC	0.822	0.917	Fully trained
MixMHCpred 2.2	0.795	0.847	Fully scored
Binding-only (MHCflurry)	0.800	0.861	Fully scored
DeepImmuno	0.698	0.710	Fully trained (9/10-mer only, n=623)

Read this honestly: BigMHC (0.822) is a near-tie with SESTRAV's canonical full_31 (0.828) and edges it on top-decile recall - but SESTRAV is scored strictly out-of-fold while BigMHC is fully trained on undisclosed data. SESTRAV's canonical full_31/mode_31 (0.828) posts the highest point AUC-PR in this certified field - a statistical near-tie with BigMHC. Source: results/table3_tier_a_metrics.csv; full methodology in paper section 3.3.

Paradigm 2 - v5 within-virus CV (N=35,597 active; 9 viruses + IEDB viral negatives + central-tolerance decoys)

The canonical same-pathogen (within-virus) discrimination metric is the per-virus within-CV table (mode-31 RF trained and evaluated per virus; results/per_virus_eval_v5_mode31.csv):

Virus	Within-CV AUC-ROC
HIV-1	0.894
DENV	0.859
IAV	0.856
CMV	0.819
EBV	0.790
HBV	0.708
SARS-CoV-2	0.699
HCV	0.575
HPV	0.561
Mean	0.751

Self-proteome Gate 1 (viral epitopes vs. self-peptide background) is a separate, valid context: AUC-PR 0.8897 (consistent with the Gate 1 threshold protocol). Both are strict 5-fold OOF. This is the model shipped for production scoring.

Note: The pooled AUC-ROC 0.9368 previously reported as a same-pathogen figure was decoy-inflated (it only reproduces when synthetic / cross-pathogen decoys, incl. the vaccinia panel, are mixed in as if they were same-pathogen negatives) and is RETRACTED. The honest pooled same-pathogen ROC on real IEDB negatives is 0.712; the pooled same-pathogen AUC-PR is base-rate-inflated (about 81% positive) and is not reported as a headline. The per-virus within-CV AUC-ROC above (mean 0.751), not any pooled number, is the reported same-pathogen metric. DENV 0.859 is itself decoy-inflated (real-negative-only ROC 0.491 on 12 negatives).

Primary metric: AUC-PR (class-imbalanced data; random baseline ≈ positive prevalence). ISSR@10 = fraction of true positives ranked in the top 10%.

Paradigm 3 - Leave-One-Virus-Out (LOO) Cross-Virus Transfer (Amendment 7)

Cross-virus transfer was evaluated by holding out each of the 9 viruses entirely from training, then testing on its IEDB-confirmed positives and IEDB assay-confirmed negatives only. An earlier analysis included synthetic decoy rows (allele_matched_nonbinder) in the test partition; because RF mode-31 binding features trivially discriminate these decoys, those figures were inflated by approximately 0.25-0.50 AUC-ROC. Amendment 7 restricts the test partition to rows where negative_origin in {tested_negative, iedb_api}. The corrected results are the ones reported here and in results/loo_cross_virus_v5_clean.csv.

Virus	LOO AUC-ROC	Within-CV AUC-ROC	n_test_pos	n_test_neg	Note
CMV	0.633	0.819	740	272	Modest transfer
HBV	0.556	0.708	325	229	Modest transfer
HCV	0.528	0.575	333	320	Marginal
EBV	0.496	0.790	316	80	Near chance
IAV	0.488	0.856	342	119	Near chance
HPV	0.468	0.561	186	137	Near chance
SARS-CoV-2	0.462	0.699	2473	980	Near chance
DENV	0.372	0.859	806	12	Unreliable (only 12 clean negatives)
HIV-1	0.162	0.894	2516	60	Anti-predictive (binding-feature reversal)
Mean	0.463	0.751			3/9 viruses above chance

Interpretation: Mean LOO AUC-ROC 0.463 indicates that mode-31 binding-derived features do not transfer reliably across viral families when tested fairly. This is an expected finding given the model's design: SESTRAV is engineered for within-virus epitope prioritization (within-CV mean AUC-ROC 0.751) and self-proteome discrimination (Gate 1 AUC-PR 0.8897). The LOO analysis characterizes the boundary of current applicability and motivates the GNN research track, where structural embeddings (ESM-2 + GINEConv) may provide more transferable representations.

SHAP Feature Attribution

SHAP analysis (Random Forest, 720 samples) attributes the decision to:

• 60% MHC binding features (per-allele presentation scores)
• 40% TCR-contact physicochemical features (positions p4-p8)

This 60/40 split confirms that TCR features provide meaningful independent signal beyond binding alone, consistent with the 9/10 gold-standard negative discrimination result.

See results/external_benchmark_comparison.md for full methodology and results/shap_values_rf.csv for per-feature SHAP values.

Pipeline Overview

SESTRAV proceeds through six computational stages under a reproducible Snakemake DAG:

graph LR
    A("Viral Proteome FASTA<br/>CMV · EBV · HBV · HCV · HPV<br/>HIV-1 · IAV · DENV · SARS-CoV-2") -->|Stage 1| B("Peptide Generation<br/>8-11mer sliding window")
    B -->|Stage 2| C("MHC Binding Prediction<br/>MHCflurry · 10-allele panel")
    C -->|Stage 3| D("TCR Feature Extraction<br/>20 physico · 10 binding · 1 length")
    D -->|Stage 4| E("Immunogenicity Scoring<br/>RF · XGBoost ensemble")
    E -.->|Stage 5 optional| F("Antigen Processing<br/>NetChop · TAPreg")
    E -.->|Stage 6 optional| G("GNN Structural<br/>Benchmark")
    E --> H("Ranked Output<br/>+ SHAP · freeze-mode governed")
    F --> H
    G --> H

Loading

1. Peptide Generation: Sliding-window extraction of 8-11mer peptides from viral proteome FASTA files.
2. MHC Binding Prediction: MHCflurry 2.2.1 presentation scores across 10 common HLA alleles (pinned version, CI-gated).
3. TCR Feature Extraction: 20 physicochemical properties at TCR-contact positions p4-p8 + 10 binding scores + peptide length = 31 features (canonical) or 33 with antigen processing tier.
4. Immunogenicity Scoring: Ensemble classification (RF, XGBoost) with SHAP interpretability and conformal prediction intervals.
5. Antigen Processing (optional, feature_mode=33): NetChop 3.1 proteasomal cleavage + TAPreg TAP transport scores as additional training features.
6. GNN Structural Benchmark (optional, research track): Graph neural network scoring (GINEConv + ESM-2) over the peptide residue graph; passes 4 of 5 promotion gates on v4 data (v5 retraining pending; ESM-2 cache complete at 27,376 peptides). See ARCHITECTURE.md for the gating policy that governs promotion to a canonical scorer.

Input: Viral proteome FASTA files (default: HPV16/18, EBV B95-8, HBV ayw, HCV 1a panels). Output: Ranked epitope candidates with immunogenicity scores, SHAP values, and visualizations.

Scope note: TCR contact positions p4-p8 follow Chowell et al. (2015), applied as a length-agnostic approximation. For 8-mer peptides, p7/p8 are zero-imputed to reflect the compressed binding register; predictions for non-canonical binding registers carry additional uncertainty.

Input Data and Naming Conventions

SESTRAV runs on bundled repository data by default. User-uploaded files are unnecessary unless intentionally overriding defaults.

Proteome Identifiers

Proteome ID	Virus	Strain(s)	Antigens	FASTA File
HPV16_18_panel8	Human Papillomavirus	HPV-16, HPV-18	8 (E2, E5, E6, E7 from each strain)	data/proteomes/HPV16_18_panel8.fasta
EBV_B95_8_panel8	Epstein-Barr Virus	B95-8	8 (EBNA1, EBNA3A, EBNA3B, LMP1, LMP2A, gp350, BZLF1, BRLF1)	data/proteomes/EBV_B95_8_panel8.fasta
HBV_ayw_panel4	Hepatitis B Virus	genotype D/ayw	4 (HBcAg, HBx, HBsAg-S, HBpol)	data/proteomes/HBV_ayw_panel4.fasta
HCV_1a_panel4	Hepatitis C Virus	genotype 1a/1b	4 (Core, NS3, NS5A, NS5B)	data/proteomes/HCV_1a_panel4.fasta

Full UniProt accessions are available in docs/antigen_accessions.md.

Output File Naming

Per-proteome outputs follow the pattern results/{proteome_id}_{suffix}:

Suffix	Contents
_peptides.csv	All 8-11mer peptides (Stage 1)
_binding.csv	MHCflurry presentation scores (Stage 2)
_features.csv	31 features per peptide (Stage 3, canonical)
_ranked.csv	Final scored and ranked epitope candidates (Stage 4)
_top20_immunogenicity.png	Bar chart of top 20 predicted immunogenic peptides
_score_distribution.png	Histogram of score distribution across all peptides

Validation and analysis outputs (committed) are summarized in results/; see the repository for the complete list.

Feature Schemas

At each TCR contact position, SESTRAV computes the following physicochemical properties. Unless otherwise referenced, properties are based on canonical amino acid physicochemical classifications.

Property	Scale / Definition	Source
Hydrophobicity	Kyte-Doolittle (-4.5 to +4.5)	Kyte & Doolittle, 1982
Aromaticity	Binary (F, W, Y, H = 1)	Canonical
Van der Waals volume	ų	Zamyatnin, 1972
Charge at pH 7	K/R = +1, D/E = -1, others = 0	Canonical
Flexibility	Vihinen flexibility (0.904 - 1.102)	Vihinen et al., 1994
Bulkiness	Zimmerman bulkiness (3.4 - 21.67)	Zimmerman et al., 1968
Hydrophilicity	Hopp-Woods (-3.4 to 3.0)	Hopp & Woods, 1981
TCR upward probability	Heuristic derived from structural alignments	Internal structural mapping

Track Definitions

Track	Features	AUC-PR (v3 OOF)	Use Case
Canonical (31-feature)	20 physicochemical + 10 binding + length	0.864	Default release track
Extended (33-feature)	31 + NetChop + TAPreg	0.886 (unweighted) / 0.840 (weighted)	Antigen processing tier - best v3 result
Legacy (30-feature)	20 physicochemical + 10 binding	0.825	Historical comparator
Legacy (21-feature)	Sequence-only (binding excluded)	0.772	Historical comparator
Expanded (50-feature)	40 physicochemical + 10 binding	-	Extended evaluation
Allele-aware (166)	Canonical + 136 HLA pocket pseudo-sequences	-	Pan-allele modeling

Stage 4 auto-detects the appropriate feature set for each trained model.

Biological Data Limitations & Mitigation

The input training data for SESTRAV contains severe biological biases inherent to public datasets (like IEDB). A quantitative breakdown of these taxonomic and topological skews is detailed in the data bias audit (internal document; key findings summarized below).

• Taxonomic skew: EBV 68.13%, HPV16 30.88%, HPV11 1.00%.
• Length skew: 9-mer peptides 64.74%.

To prevent machine learning models from over-indexing on EBV-specific anchor motifs and 9-mer length preferences (which would lead to poor generalization on minority taxa like HPV11 or non-canonical peptide lengths), the compute_sample_weights() function in features.py is CRITICAL. It dynamically calculates sample weights to up-weight minority taxa and non-9-mer peptides during model training, balancing the learning signal and ensuring robust pan-viral performance.

Quick Start

1. Environment Setup

Conda (recommended for reproducibility):

conda env create -f environment.yml
conda activate sestrav
mhcflurry-downloads fetch models_class1_presentation

pip (editable/dev):

pip install -e ".[dev]"          # lint + test tools
mhcflurry-downloads fetch models_class1_presentation

venv:

python -m venv .venv
source .venv/bin/activate      # Linux/macOS
pip install -r requirements.txt
pip install snakemake
mhcflurry-downloads fetch models_class1_presentation

2. Install and Train Models

# Core install from source (not yet published to PyPI)
git clone https://github.com/Gavin-Borges/SESTRAV.git
cd SESTRAV
pip install .

# With GNN structural scorer
pip install ".[gnn]"

# With Snakemake pipeline runner
pip install ".[pipeline]"

# Developer install (ruff, mypy, pytest)
pip install -e ".[dev]"

Models must be trained before production pipeline execution:

# Canonical 31-feature track (recommended)
python -m src.train_classifier \
  --data data/immunogenicity_dataset_v5.csv \
  --feature-mode 31 \
  --binding-matrix models/peptide_binding_matrix_v5.csv \
  --sample-weights

# CLI equivalent
sestrav validate \
  --dataset data/immunogenicity_dataset_v5.csv \
  --feature-mode 31 \
  --binding-matrix models/peptide_binding_matrix_v5.csv \
  --sample-weights \
  --report results/validation_report_v5.md

Note: Without trained models, the pipeline falls back to a prototype mode using binding-derived pseudo-labels (for testing only; not scientifically valid).

3. Run the Pipeline

# Snakemake (recommended)
snakemake --snakefile pipeline.smk --cores 4

# Standalone entry point
python pipeline.py

4. Validate Release Readiness

git status
python -m pytest tests/test_features.py tests/test_metrics.py tests/test_pipeline_integration.py -q
snakemake --snakefile pipeline.smk --dry-run --cores 1   # optional

For freeze-grade validation:

snakemake --snakefile pipeline.smk full_validation_report --cores 4 --forceall

5. Post-Pipeline Analysis (Optional)

Generate full validation report:

snakemake --snakefile pipeline.smk full_validation_report --cores 4

Prepare inputs for external tool comparison (PredIG, PRIME):

python -m src.prepare_external_validation_inputs
python -m src.external_benchmark_comparison --predig ... --prime ...

See scripts/README.md for the external-validation utilities and workflow.

6. ANN / GNN Benchmarks (Optional)

• ANN: pip install -r requirements-ann.txt, then python -m src.ann_benchmark --help. Default architecture: 256-128-64 ReLU, dropout 0.2 (AUC-PR = 0.8252 ± 0.0248).
• GNN v2.3 (research track): pip install ".[gnn]" (from source; not published to PyPI), then python -m src.train_gnn --help. Architecture: GINEConv x2 over a per-residue peptide graph with ESM-2 node embeddings (facebook/esm2_t12_35M_UR50D, 480-dim), fused with the canonical mode-31 physicochemical features. On v4 data it passes 4 of 5 promotion gates (mean-fold AUC-PR 0.7281); v5 retraining is in progress (ESM-2 cache complete at 27,376 peptides; GPU training pending). The GNN remains a research track pending v5 evaluation. Pre-compute ESM-2 embeddings with scripts/precompute_esm2_embeddings.py before training.

7. Google Colab

A Colab-ready script is available in notebooks/SESTRAV_Colab_Pipeline.py; see notebooks/README.md for details.

Container Quick Start

The Docker image does not include trained models. Build and then train:

docker build -t sestrav:latest .
docker run --rm -v "$(pwd)/models:/app/models" sestrav:latest \
  -m src.train_classifier --data data/immunogenicity_dataset_v5.csv \
  --feature-mode 31 --binding-matrix models/peptide_binding_matrix_v5.csv

Run the pipeline with bind-mounted directories:

mkdir -p results
docker run --rm \
  -v "$(pwd)/models:/app/models" \
  -v "$(pwd)/results:/app/results" \
  sestrav:latest

Windows PowerShell:

New-Item -ItemType Directory -Force results | Out-Null
docker run --rm `
  -v "${PWD}/models:/app/models" `
  -v "${PWD}/results:/app/results" `
  sestrav:latest

Container smoke test (recommended before release):

docker run --rm -v "$(pwd)/data:/app/data:ro" sestrav:latest -m pytest tests/ -q --basetemp=tmp_pytest

API & Demo Quick Start (Docker Compose)

A two-service Docker Compose stack serves the FastAPI microservice and Streamlit demo from pre-trained model artifacts. Model binaries must be present in models/ before launching.

# Build and launch both services
docker compose up --build

# FastAPI docs:  http://localhost:8000/docs
# Streamlit demo: http://localhost:8501

Single-peptide API request:

curl -X POST "http://localhost:8000/score" \
  -H "Content-Type: application/json" \
  -d '{"sequence":"GILGFVFTL","allele":"HLA-A*02:01"}'

Both services bind to 127.0.0.1 only (loopback) to prevent unintended public exposure on shared research machines. Model artifacts are mounted read-only.

Virus	Antigens
EBV (8)	EBNA1, EBNA3A, EBNA3B, LMP1, LMP2A, gp350, BZLF1, BRLF1
HPV (8)	HPV16 E2, E5, E6, E7; HPV18 E2, E5, E6, E7

Evaluation Metrics

All metrics are computed by src/evaluate_metrics.py.

Metric	Description
AUC-PR	Area Under Precision-Recall Curve (primary metric, robust to class imbalance)
AUC-ROC	Area Under ROC Curve
ISSR@10/25	True positive fraction in top 10% / 25% (enrichment)
Precision@10/25	Precision among top 10% / 25% predictions
Recall@10/25	Recall captured in top 10% / 25%
NDCG@10/25	Normalized Discounted Cumulative Gain at top 10% / 25%

Reproducibility and Data Provenance

Included in this repository:

• Training dataset (data/immunogenicity_dataset_v5.csv, 35,597 active rows / 51,185 total; v4 retained for historical comparison)
• Viral proteomes (data/proteomes/)
• Binding matrix (models/peptide_binding_matrix_v5.csv) and model metadata
• All pipeline code, tests, and documentation

Generated locally (excluded from git):

• Trained model binaries (*.joblib, *.pt)
• Most workflow outputs in results/ (except committed validation snapshots)
• Runtime caches

A fresh clone must run model training before production scoring. Release bundles with SHA256 manifests can be created via python -m src.release_bundle.

Training labels are derived from curated IEDB-linked immunogenicity evidence. Publications should cite both this repository and the original upstream data sources.

Documentation

Document	Description
docs/feature_glossary.md	Feature definitions and track schemas
docs/antigen_accessions.md	Full UniProt accessions and gene names
docs/output_naming_standard_v1.md	Output file naming policy
src/naming.py	Legacy proteome/model ID alias compatibility
docs/validation_summary.md	Detailed validation results and interpretation
docs/limitations_statement_v1.md	Standardized external communication language

Cite This Work

If you use SESTRAV in your research, please cite this repository:

@software{borges2026sestrav,
  author    = {Borges, Gavin and Eljamal, Abdelrahman and Schellenberg, Iris and
               Jouaneh, Charles and Byers, Emine},
  title     = {{SESTRAV}: Structural Epitope Scoring via {TCR} Recognition And Vaccinology},
  year      = {2026},
  url       = {https://github.com/Gavin-Borges/SESTRAV},
  version   = {2.0.3}
}

See CITATION.cff for the full machine-readable citation.

License

MIT License. See LICENSE for details.

Maintainers and Contributors

Lead Developer & Maintainer (SESTRAV 2.0)

• Gavin Borges

Original SESTRAV 1.0 Foundation Team (University of Rhode Island)

• Abdelrahman Eljamal: ML Engineer & Computational Architect
• Iris Schellenberg: Translational Vaccine Strategy, Data Finding, and Curation
• Charles Jouaneh: Vaccine Strategy & Bioinformatic Pipeline Development
• Emine Byers: Structural Immunology & Data Curation

Developed by Gavin Borges. Academic acknowledgements: bioinformatics coursework at NC State (BPS 542 / CMB 522 / CSC 522 / STA 522; CMB 523) provided foundational grounding; SESTRAV is an independently maintained research tool.