BIO-SEMANTICS

Gene-Expression Semantic Modeling Pipeline
AHMAD ALI PARR · JESSICA L. WILLIAMS · SNAPKITTYWEST
Evidence or Silence · 2026

cDNA Observations

OBSERVED
GENESAMPLECOUNTNORM
BRCA1S0014,8928.24
TP53S00112,34011.02
MYCS0028916.15
EGFRS002MISSINGMISSING
KRASS0032,1077.89
PTENS0036,7239.41
6 observations · 3 samples · 6 genes · 1 missing cell

Gene Expression Matrix

GENE
S001
S002
S003
BRCA1
8.24
5.12
7.01
TP53
11.02
9.88
10.45
MYC
4.21
6.15
3.89

Data Validation

semantics.py · 10 VALIDATOR RULES
01Version and discriminant check 02Value XOR missing reason 03Canonical unsigned decimal counts 04Finite decimals with units 05Sample/gene reference resolution 06Quantity kind integrity 07Exclusion map preservation 08Source byte completeness 09Synthetic origin propagation 10No phantom model results

Architecture Pipeline

BIOLOGICAL OBSERVATIONS
cDNA EXPRESSION DATA
DATA VALIDATION
GENE EXPRESSION
STAN MODEL
POSTERIOR STATE
MINIZINC MODEL
COMPUTATIONAL RESULT
VISUALIZER
HUMAN-READABLE MODEL

Semantic Quantity Classification

OBSERVED — measurement INFERRED — posterior DERIVED — transformation PREDICTED — model output CONSTRAINED — solver ENCODED — representation EXECUTED — run evidence
Observed measurements ≠ inferred posteriors.
Inferred posteriors ≠ deterministic derivations.
Derivations ≠ binary representations.
Representations ≠ execution evidence.

PTM — Bit Parity / 1

q_even q_odd q_accept
1
0
1
1
0
1
0
_
STATEREADWRITEMOVENEXT
q_even00Rq_even
q_even11Rq_odd
q_odd00Rq_odd
q_odd11Rq_even
q_even_0Stayq_accept
q_odd_1Stayq_accept
n INPUT SYMBOLS · n+1 TRANSITIONS · O(1) AUXILIARY STORAGE

Stan — Bayesian Inference

INFERRED

Prior → Likelihood → Posterior

baseline
6.84
cond_eff
1.23
time_eff
0.47
sigma
2.11
PARAMMEANSD95% CIR-HATESS
baseline6.840.42[6.02, 7.66]1.004200
cond_eff1.230.31[0.63, 1.84]1.003800
time_eff0.470.18[0.12, 0.82]1.012900
sigma2.110.15[1.82, 2.41]1.004100
0 divergences · run_id: not_run

MiniZinc — Constraints

CONSTRAINED
VARIABLEDOMAINCONSTRAINTSTATUS
sample_assign{0,1}balanced groupsfeasible
batch_block1..4no confoundingfeasible
gene_select{0,1}top-k variancenot_run
solver_outcome: not_run

Quipper — Quantum Circuit

q0: ─── H ──── ● ──── M ─── q1: ──────── X ──── M ─── q2: ─── H ──────── M ─── DEPTH: 3 QUBITS: 3 GATES: 4 MEASUREMENTS: 3 ANCILLAS: 0
generation: specified · validation: not_run · execution: not_run

Binary Semantics

ENCODED
PROFILE: u32-checked/1
unsigned 32-bit · big-endian · 0..4294967295
overflow: trap · div-by-zero: trap

State Transition

SEMANTIC VALUE → ENCODING → BITS → STATE → TRANSITION
00000000 00000000 00010011 10110010
present zero ≠ missing · distinct encodings

Immutable Provenance

RPG / JCL Integration

WORKFLOW: RUNSTAGE
STAGE: PYTHON_PIPELINE
STATUS: NOT_RUN
//VAULTJOB JOB (ACCOUNT)
DSN=YOURHLQ.VAULT.RUN0001.REPORT
DSN=YOURHLQ.VAULT.RUN0001.MANIFEST
STATUS: TEMPLATE ONLY

Repository Structure

bio-semantics/
  src/
    semantics.py
    test_semantics.py
    dashboard.html
    index.html
  spec/
    types.ts
    contract.md
    system_prompt.md
  templates/
    workflow.rpgle
    vault.jcl

Gated Phases

GATESTATUSEVIDENCE
1. CONTRACTREADYContracts specified, validators written
2. ENGINENOT STARTEDStan/MiniZinc/Quipper sources pending
3. INTEGRATIONNOT STARTEDPanels, WORM, governance pending