Full-sequence context audit

Evidence inventory, not a ranking or clinical safety verdict. Target masks, MHC predictions, cleavage observations, independently documented intended ligands and historical selection agreement are different facts. This report overlays masks and all returned MHC predictions; it does not establish intended-ligand identity, recognition or presentation.

All intervals are zero-based, half-open [start,end). Bond b separates residues b−1 and b. Sequence endpoints 0 and N are not peptide bonds. A native context may be a cropped RNA-backed window, not a complete protein or an exposed molecule. A translated product is the complete CDS translation, not necessarily its mature processed form.

Self-source/CTA policy and near-self assessment are not performed here. Use attributed construct safety assessments separately; unassessed self evidence is not absence of self ligands. No serum half-life, clearance, localization or intact-delivery conclusion is made.

{{ ctx.name }} — {{ ctx.kind }}

{{ ctx.sequence|length }} aa; identity {{ ctx.source_id }}

{{ ctx.sequence }}

Terminal chemistry: N={{ ctx.n_term }}, C={{ ctx.c_term }}. Basis: {{ ctx.chemistry_basis or 'not established' }}. Model assumptions never replace manufacture provenance.

Target mapping: {{ ctx.target_mapping_status }}. {% if ctx.construct %}Manufacture evidence: {{ ctx.construct.evidence }}. Native window: [{{ ctx.construct.native_start }},{{ ctx.construct.native_end }}). Native residue offsets (None = added/replaced/unresolved): {{ ctx.construct.native_residue_offsets }}. Edits: {{ ctx.construct.sequence_edits }}. Chemical modifications: {{ ctx.construct.chemical_modifications }}. Removed target intervals: {{ ctx.construct.removed_target_intervals }}.{% endif %}

{% if ctx.product %}

Complete CDS and emitted nucleotide views validated. Evidence: {{ ctx.evidence }}.

Complete translated product and nucleotide provenance

CDS nt: {{ ctx.product.cds_nt }}

Full nt: {{ ctx.product.full_nt }}

Elements: {{ ctx.product.elements }}

All construct placements: {{ ctx.product.construct_placements }}

{% endif %}

Source provenance — every occurrence retained

{% for source in ctx.source_antigens %} {% else %}{% endfor %}

Kind	Gene / ID	Transcripts / proteins	Species / source
{{ source.kind }}	{{ source.gene_name }} / {{ source.gene_id }}	{{ source.transcript_ids\|join(', ') }} / {{ source.protein_ids\|join(', ') }}	{{ source.species }} / {{ source.source_identifier }}
Source provenance unassessed

Target masks

{% for target in ctx.targets %} {% else %}{% endfor %}

ID	Source	Interval	Sequence / junction
T{{ loop.index }}	{{ target.source_name }}	[{{ target.start }},{{ target.end }})	`{{ ctx.sequence[target.start:target.end] or 'zero-width target junction' }}`
No mapped target intervals; see mapping status above.

Patient-MHC inventory: {{ audit.mhc_status }}

Coverage refers only to the explicit requests below, not all patient alleles or all models. Unrequested class II or other combinations remain unassessed. A returned row is not necessarily a strong binder. All output is retained without ranking or immunogenicity filters.

{% for req, observed, missing in audit.coverage %} {% else %}{% endfor %}

Kind / model / version	Allele / length	Observed offsets	Missing offsets
{{ req.kind }} / {{ req.predictor_name }} / {{ req.predictor_version }}	{{ req.allele }} / {{ req.peptide_length }}	{{ observed }}	{{ missing }}{% if req.peptide_length > ctx.sequence\|length %} (no fitting windows){% endif %}
Requested MHC coverage unassessed

{% if audit.unrequested_predictions %}

{{ audit.unrequested_predictions|length }} unrequested model/HLA/length outputs retained; check request provenance.

{% endif %} {% for ligand in audit.ligands %} {% else %}{% endfor %}

ID / interval / peptide	Target-mask overlaps	Every model observation (native scales)
L{{ loop.index }} [{{ ligand.start }},{{ ligand.end }}) `{{ ligand.peptide }}`	{% for target in ctx.targets %}{% if target.overlaps(ligand.start,ligand.end) %}T{{ loop.index }} {% endif %}{% endfor %}	{% for pred in ligand.predictions %}{{ pred.kind }} / {{ pred.allele }} / {{ pred.predictor_name }} {{ pred.predictor_version }}: score={{ pred.score }}, value={{ pred.value }} {{ value_unit(pred.kind) or '(unitless/no linear unit)' }}, percentile={{ pred.percentile_rank }} {% endfor %}
No ligand observations; see status and missing coverage above.

Full-sequence processing evidence

Proteasomal, other cytosolic, ER, endolysosomal and extracellular/serum evidence remain separate. Only the model/bond observations below were assessed. Other models and compartments remain unassessed. An enzyme compartment annotation does not establish matrix calibration or delivery into that compartment. Motif non-match is not resistance; native scores are not interchangeable.

{% for view in record.profiles %}{% set p=view.profile %}

{{ p.model.name }} {{ p.model.version }} — {{ p.status }}

Enzyme {{ p.model.enzyme }} / UniProt {{ p.model.uniprot }} / species {{ p.model.species }} / compartments {{ p.model.compartments }}. Evidence: {{ p.model.evidence }}. Assay: {{ p.model.assay }}.

Score: {{ p.model.score_name or 'qualitative observation only' }}; units: {{ p.model.score_units or 'none' }}. Limitations: {{ p.model.limitations }}. References: {{ p.model.references }}.

Model input: N={{ p.peptide.n_term }}, C={{ p.peptide.c_term }}. Settings {{ p.settings }}; backend {{ p.backend_version }}; model-asset SHA-256 {{ p.model_asset_sha256 or 'not supplied' }}.

Raw terminal output: {{ p.terminal_output }} (endpoint sentinel, never a bond or a cleavage probability). Raw residue vector: {{ p.raw_residue_scores }}.

{% for b, roles in view.roles.items() %}{% set site=view.sites.get(b) %}{% endfor %}

Bond / residues	Observation / reason / native score	Padded context	Target masks: internal / boundary	Ligands: internal / N-boundary / C-boundary
{{ b }} `{{ ctx.sequence[b-1] }}\|{{ ctx.sequence[b] }}`	{% if site %}{{ site.status }} / {{ site.reason }} / {{ site.score }}{% else %}unassessed{% endif %}	{{ b in p.padded_bonds }}	{{ roles.target_internal }} / {{ roles.target_boundary }}	{{ roles.ligand_internal }} / {{ roles.ligand_n }} / {{ roles.ligand_c }}

{% else %}

No processing models requested; every compartment is unassessed.

{% endfor %}