Biological Uncertainty · GRCh38 Source Layer 1

Unknown biology is not bad evidence. Untrusted evidence is not biology.

A public HIR/OAM map for genomic uncertainty, GRCh38 reference limits, provenance gates, measurement failure, and hard blocks against hidden positive evidence.

Read invariant Explore source model Parent hub

Non-negotiable invariant

Unknown biology may preserve possibility space. Untrusted data may invalidate the comparison. Neither may be converted into hidden positive evidence.

Biological unresolvedness ≠ measurement/provenance unresolvedness. These categories must remain separate throughout all schemas, scoring, and downstream layers.

GRCh38 first-pass source model

GRCh38 is treated as Source Layer 1: an initial baseline reference frame, not a complete model of human diversity and not a final biological authority.

🧬
Layer 1

Reference frame

GRCh38 / GRCh38.p14 anchors first-pass coordinates and feature typing while carrying explicit reference limitations.

GRCh38reference
⚠️
Limits

Known gaps

Gap-adjacent, centromeric, telomeric, repeat-dense, and segmentally duplicated regions must carry reference confidence limits.

gapsconfidence
🧾
Provenance

Evidence chain

Contamination, mixed sample, broken chain of custody, and undocumented provenance invalidate or suspend comparison authority.

provenancechain
🔬
Feature

Typed features

SNPs, indels, structural placeholders, gap regions, regulatory placeholders, and unknown unresolved features require explicit status.

schemafeature classes
⛔
Hard stop

No hidden positives

Unknown biological or measurement/provenance classes may not be promoted into positive similarity, identity, or continuity evidence.

blockedHIR
🧭
Next

Staged layers

Layer expansion should proceed through T2T-CHM13, pangenome references, and functional annotation only after schema validation.

T2Tpangenome

HIR rule layer

Honesty

Every feature record must carry explicit status, uncertainty class, reference confidence, and coverage confidence.

Unknown must be labeled unknown. No field may be implicitly clean. Not assessed cannot default to high confidence.

Integrity + Respect

Biological unresolvedness and measurement/provenance unresolvedness remain separate; identity and family-relation claims are blocked beyond declared evidence limits.

Category A ≠ Category B. Biological unknown ≠ sample failure. Sample failure ≠ biology. No identity/family/continuity inference.

Staged ingest discipline

Do not ingest raw genome files or perform scoring until schema, uncertainty classes, and provenance rules are validated with sample records.

1. GRCh38 / GRCh38.p14 — active Source Layer 1 2. Machine-readable HIR rule table 3. Structured GRCh38 gap-region catalog 4. Sample feature records in JSONL 5. T2T-CHM13 source model 6. Liftover / coordinate alignment spec 7. Pangenome graph-coordinate extension before HPRC Layer 3

Boundary

This is not a genomic analysis pipeline or medical/forensic authority.

This Space is a bounded architecture and public review prototype. It is not clinical advice, diagnosis, treatment guidance, identity proof, family-relation proof, forensic conclusion, ancestry result, genetic counseling, or validated genomic-comparison software.

GRCh38 is used as a reference frame only. All genomic inferences beyond declared evidence limits are explicitly blocked by the HIR rule set.

Structural correspondence, not ontological equivalence.