An evidence inventory, not a safety verdict. A high model score does not prove a target antigen will be destroyed. A motif non-match does not establish resistance. Unreturned bonds are unassessed, never assigned zero. Enzyme location does not establish calibration in serum or another compartment. Cleavage evidence is not whole-peptide half-life, presentation or T-cell recognition.
{% if not records %}No sequence profiles supplied; processing is unassessed.
{% endif %} {% for record in records %}{% set p = record.profile %}{% set n = p.peptide.sequence|length %}Exact input ({{ n }} residues): {{ p.peptide.sequence }}
Source start: {{ p.peptide.source_start }} (zero-based). N terminus: {{ p.peptide.n_term }}; C terminus: {{ p.peptide.c_term }}. Chemistry describes the supplied model input, not independently verified manufacture.
Status: {{ p.status }}. {{ p.reason_codes|join('; ') }} {{ p.error_message }}
Enzyme: {{ p.model.enzyme }} {{ p.model.uniprot }}. Species: {{ p.model.species or 'Not specified' }}. Compartments: {{ p.model.compartments|join(', ') }}. Evidence: {{ p.model.evidence }}.
Model: {{ p.model.name }} {{ p.model.version }}; mhctools: {{ p.backend_version or 'Unknown' }}.
Score: {{ p.model.score_name or 'No numeric score' }};
native units: {{ p.model.score_units or 'Not applicable' }}.
Asset SHA-256: {{ p.model_asset_sha256 or 'Not recorded' }}.
Benchmark endpoint the native score answers: {{ p.model.scored_endpoint }}. The score is a model output for that measurement type, not a measurement.
{% endif %} {% if p.model.motif_strictness %}Motif strictness: {{ p.model.motif_strictness }} — {% if p.model.motif_strictness == 'required' %}the source establishes this pattern as necessary for the activity, so a non-match is meaningful evidence against cleavage by this enzyme through this route, and still not proof of resistance. {% elif p.model.motif_strictness == 'preferred' %}the source reports a favoured context; non-matching bonds can still be cleaved, usually more slowly, so a non-match is weak evidence. {% else %}a broad flag with many exceptions: a match constrains little and a non-match almost nothing. {% endif %} Grade basis: {{ p.model.strictness_basis or 'Not recorded' }}.
{% endif %}Settings: {% for key, value in p.settings %}{{ key }}={{ value }}; {% endfor %}
Assay/training: {{ p.model.assay }}. Limitations: {{ p.model.limitations }}
Sources: {% for source in p.model.references %}{{ source }}
{% endfor %}
Target masks and patient-HLA ligand overlays are not included in this low-level site inventory. Other models and processing compartments remain unassessed unless explicitly reported. Padded context is missing sequence context, not measured flanking sequence. A complete vector is not complete biological coverage.
| Input bond b | Source bond | Residues | Status | Native score | Context | Reason |
|---|---|---|---|---|---|---|
| {{ b }} | {{ p.peptide.source_start + b }} | {{ p.peptide.sequence[b - 1] }}|{{ p.peptide.sequence[b] }} |
{{ site.status if site else 'unassessed' }} | {{ site.score if site and site.score is not none else '' }} | {{ 'terminal context padded' if b in record.padded else 'see model/input scope' }} | {{ site.reason if site else 'No observation returned for this bond' }} |
The sequence has no internal peptide bonds.
{% endif %}Bond b splits sequence[:b] | sequence[b:]. Positions 0 and {{ n }} are sequence endpoints, not internal peptide bonds.
{% if p.raw_residue_scores %}Complete residue-indexed backend output:
{{ p.raw_residue_scores|join(', ') }}.
Final residue output: {{ p.terminal_output }} — retained endpoint sentinel, excluded from internal cleavage evidence.
{% endif %}