Sequence is not only information. It is matter in a state.

Flux Genome Physics in Matter Computing

Sequence identifies biological matter; physical state determines what that matter can do. DNA and RNA occupy structures, environments, and interaction states that change with pairing, modification, geometry, mechanics, and molecular context.

Flux Genome Physics defines a planned extension of Matter Computing from sequence state to local properties, mutation or design deltas, molecular interactions, and experimentally testable decisions. Physical results and biological interpretations occupy separate claim levels, each with its own evidence requirements.

1. Genome Physics Builds on Existing Domains

FluxMateria already applies one physical foundation across:

• Chemistry;
• Materials;
• Pharmacology;
• molecular interaction;
• protein and docking workflows;
• high-throughput candidate search;
• provenance and validation infrastructure.

Genome Physics extends that platform upward.

It does not replace the deployed domains.

It depends on them.

Chemistry supplies the molecular foundation

Nucleotides are chemical systems.

Pairing, stacking, backbone geometry, modifications, hydration, ions, and ligand interaction depend on molecular physics.

Pharmacology supplies therapeutic context

Mutated targets, compound rescue, protein interaction, exposure, safety, and assay selection connect sequence physics to therapeutic decisions.

Materials supplies interface context

Delivery systems, sensors, substrates, nanoparticles, and assay environments connect biological matter to engineered matter.

Genome Physics is therefore the next rung of the Matter Ladder:

Atoms
    ↓
Molecules
    ↓
Chemistry
    ↓
Proteins and molecular interaction
    ↓
DNA and RNA
    ↓
Regulation
    ↓
Therapeutic sequence design

The platform is not entering an unrelated market.

It is extending one engine into a more complex organization of the same matter.

2. The Sequence-State Contract

The defining Genome Physics rule is:

Sequence identifies the matter. State determines the physical problem.

The same base sequence may behave differently depending on:

• DNA or RNA chemistry;
• strand orientation;
• pairing partner;
• protonation;
• chemical modification;
• salt and ion concentration;
• pH;
• temperature;
• solvent and hydration;
• circularity, supercoiling, or topology;
• tension, force, or confinement;
• length and boundary conditions;
• protein or ligand binding;
• assay or cellular context.

A minimal FASTA input may support a first-pass local scan.

A higher-authority calculation may require explicit:

• backbone chemistry;
• strand state;
• modification pattern;
• pairing state;
• structural state;
• environment;
• target or protein partner;
• assay conditions.

The interface must separate six layers.

Sequence identity

Which nucleotides, backbones, lesions, and modifications are present?

Strand state

Single-stranded, duplex, guide–target, hairpin, circular, linear, or another declared configuration?

Structural state

Which pairing, stacking, local geometry, topology, and competing structures are represented?

Environmental state

Which temperature, salt, pH, ions, solvent, confinement, and force apply?

Interaction state

Which protein, ligand, nuclease, polymerase, ribosome, compound, or target sequence is present?

Decision context

Which screening, therapeutic, editing, construct, or regulatory question is being asked?

A sequence-only result must declare what it assumed.

Minimal input is an interface convenience.

Missing state remains a scientific limitation.

3. One Engine, a Genome-Physics-Specific Compiler

Flux Genome Physics is planned as the nucleic-acid compiler and interface of one Matter Computing Platform.

Its modules share one nucleic-acid representation, compiler, evidence system, and physical foundation.

Flux Theory
    ↓
Versioned Foundation Interface
    ↓
Matter Graph
    ↓
Genome Physics Compiler
    ↓
Sequence state · Structure · Interaction
    ↓
Mutation delta · Candidate search · Redesign
    ↓
Genome Physics Decision Object

The interface must contribute:

• nucleotide, backbone, lesion, and modification representations;
• strand, pairing, and topology state;
• Flux-derived nucleic-acid Domain Closures;
• local structural and mechanical calculation;
• structural-state search;
• protein–nucleic-acid and ligand–nucleic-acid interaction passes;
• mutation and redesign transforms;
• therapeutic sequence search;
• provenance, evidence, and readiness logic.

It may reuse:

• atomic and bond primitives;
• molecular geometry and conformation;
• environmental and solvation infrastructure after authority review;
• protein structure and interaction workflows;
• mutation and candidate enumeration;
• staged search;
• frozen-prediction and external-validation infrastructure.

It may not introduce a second source of physical truth.

4. The Genome Physics Matter Graph

A Genome Physics problem must represent more than letters.

The Matter Graph may include:

• nucleotide identity;
• DNA or RNA sugar chemistry;
• backbone connectivity;
• strand direction;
• strand count;
• complementary or target strand;
• base pairing and stacking;
• chemical modifications;
• methylation or other physical marks;
• mismatches;
• insertions and deletions;
• lesions and damage;
• local geometry;
• helical state;
• bend and torsion;
• stems, loops, bulges, and alternative structures;
• circularity and supercoiling;
• sequence boundaries;
• temperature, salt, ions, pH, solvent, and force;
• protein, ligand, guide, nuclease, or polymerase partner;
• reference and altered states;
• construct constraints;
• therapeutic objective.

The graph separates six objects.

Identity

Which physical constituents are present?

State

How are they organized under the declared conditions?

Structural alternatives

Which competing local or secondary states are admissible?

Interaction

Which partner and physical engagement mode are being calculated?

Perturbation

Which mutation, mismatch, damage, modification, or redesign distinguishes the states?

Decision

Which candidate, mechanism, experiment, or redesign is being selected?

The Matter Graph is not only a representation format.

It is the physical contract behind every output.

5. Sources of Physical Authority

Every authoritative Genome Physics property must emerge from Flux physics.

This requirement applies before the first production line of Genome Physics code is written.

Inputs permitted in property calculation

A Genome Physics calculation may consume:

• approved outputs from the versioned Flux Theory Foundation Interface;
• Flux-derived nucleotide, duplex, RNA, interaction, and multiscale Domain Closures;
• sequence identity and chemical modifications;
• explicit structural state;
• experimental structure supplied as an input;
• measured boundary and environmental conditions;
• user constraints defining the task;
• numerical methods executing the same Flux calculation;
• non-authoritative AI and search aids that do not determine or repair a property;
• reference data used only after prediction freeze for validation or falsification.

Inputs excluded from property calculation

The following may not determine, correct, or replace a Matter Computing property:

• empirical nearest-neighbor property tables;
• fitted melting-temperature formulas;
• fitted salt or concentration corrections;
• empirically assigned stacking energies;
• lookup-derived mismatch penalties;
• motif tables used as property sources;
• fitted sequence-family coefficients;
• benchmark-optimized thresholds;
• target-specific adjustment;
• empirical residual repair;
• learned folding energies;
• learned accessibility or off-target values substituted for physical calculation;
• learned property prediction or residual correction.

A conventional table or learned model may be used as:

• comparator;
• baseline;
• candidate generator;
• search prior;
• external structure proposal;
• validation reference.

It may not become the authoritative Genome Physics property.

Disclosure does not make a prohibited ingredient acceptable.

The scientific standard is:

If the property did not emerge from Flux physics, it is not an authoritative Flux Genome Physics property.

6. What May Be Retrieved, Annotated, or Proposed

Genome Physics will legitimately use biological information.

The question is whether that information defines the task, proposes a candidate, or determines a property.

Representation and reference data

These may define the input:

• FASTA sequence;
• transcript or genome assembly;
• reference and alternate alleles;
• chemical-modification map;
• protein sequence;
• experimental structure;
• assay conditions;
• target region;
• construct specification.

Annotation and catalog data

These may provide context:

• gene and transcript annotation;
• promoter, exon, intron, UTR, and splice annotation;
• known variant catalog;
• protein and nuclease catalog;
• PAM and targeting constraints;
• oligo chemistry catalog;
• manufacturing constraints;
• delivery context;
• off-target sequence panel.

Candidate-generation data

These may propose candidates:

• sequence windows;
• guide enumeration;
• oligo libraries;
• synonymous variants;
• promoter or UTR variants;
• compound libraries;
• AI-generated sequences;
• motif-preserving mutation sets.

Validation data

These may evaluate frozen outputs:

• duplex geometry;
• melting or hybridization measurements;
• mismatch measurements;
• mechanical measurements;
• folding assays;
• binding data;
• cleavage data;
• reporter assays;
• construct stability;
• resistance measurements.

Flux-derived physical output

If a result is described as Matter Computing, quantities such as the following must come from the Flux calculation:

• pairing energy;
• stacking energy;
• local geometry;
• torsional or bending stiffness;
• opening tendency;
• mismatch delta;
• strand-interaction energy;
• structural-state energy;
• protein–nucleic-acid interaction quantity;
• ligand–nucleic-acid interaction quantity;
• mutation-induced physical delta.

Annotation may locate a promoter.

It may not determine accessibility.

A folding tool may propose a state.

It may not determine the authoritative energy.

AI may generate a sequence.

It may not provide the property used to promote it.

7. The Genome Physics Matter Program

A Matter Program defines the complete physical task.

Local sequence-physics program

System:
    DNA or RNA sequence

State:
    strand type
    pairing state
    modifications
    temperature, salt, pH, solvent

Requested outputs:
    local geometry
    pairing and stacking
    mechanics
    opening tendency
    state alternatives and warnings

Mutation-delta program

System:
    reference sequence or target state

Perturbation:
    substitution, mismatch, insertion, deletion, lesion, or modification

Objective:
    calculate the physical delta

Output:
    changed properties
    affected structure or interaction
    mechanism trace
    uncertainty
    decisive experiment

Oligo-triage program

Target:
    declared transcript or genomic region

Candidates:
    ASO, siRNA, guide, primer, or probe sequences

Constraints:
    length
    chemistry
    target window
    off-target panel
    manufacturability

Output:
    ranked candidates
    self-structure risk
    target interaction
    mismatch discrimination
    backup families

Construct program

System:
    mRNA, plasmid, vector payload, DNA vaccine, or synthetic construct

Constraints:
    encoded product
    required motifs
    manufacturability
    modification policy

Output:
    physical-risk map
    structural hotspots
    redesign candidates
    preserved-function report

Resistance-rescue program

System:
    reference and mutated therapeutic target

Candidates:
    compound or sequence library

Objective:
    retain interaction under mutation

Output:
    target-state delta
    retained candidates
    lost interactions
    rescue hypotheses
    frozen experimental shortlist

The program is the contract between biological intent and physical computation.

8. The Genome Physics Compiler

The Compiler transforms the program into an executable sequence-to-decision cascade.

Parse

Read sequence, modifications, strands, target, environment, constraints, and objective.

Normalize

Resolve alphabet, orientation, complement, modifications, units, numbering, and reference state.

Validate

Reject incomplete, unsupported, unsafe, or physically inconsistent specifications.

Classify the output

Select the Foundation Interface and approved Flux-derived Genome Physics Domain Closures.

Build state

Construct the nucleic-acid Matter Graph under declared conditions.

Enumerate structures or candidates

Generate admissible pairing, geometry, secondary-structure, sequence, or compound candidates.

Generation is non-authoritative.

Calculate

Evaluate authoritative Flux properties for every promoted state.

Resolve interaction

Add target sequence, protein, ligand, guide, nuclease, polymerase, or other declared partner.

Apply perturbation

Construct reference, mutation, damage, modification, or redesign states.

Search

Generate, reject, and rank candidates under physical and user constraints.

Assemble

Return a Genome Physics Decision Object containing:

• authoritative properties;
• generated and inferred states;
• mutation or design deltas;
• candidate ranking;
• rejected candidates and reasons;
• biological hypotheses, separately labeled;
• uncertainty and applicability;
• evidence and provenance;
• recommended experiment.

The Compiler must preserve the distinction:

Generate a state
        ≠
Calculate its property
        ≠
Infer a biological consequence

9. The Physical Claim Ladder

Genome Physics must prevent local physical evidence from silently becoming a biological or clinical claim.

The claim ladder has six levels.

P0 — State definition

What sequence, structure, environment, and partner were represented?

P1 — Local physical property

What geometry, energy, mechanics, opening, or stability quantity was calculated?

P2 — Interaction or perturbation delta

What changed between states, candidates, mismatches, or mutations?

P3 — Mechanism hypothesis

Which physical mechanism could explain an observed or proposed biological effect?

P4 — Biological hypothesis

What expression, editing, regulation, activity, or phenotype may follow?

P5 — Therapeutic or clinical outcome

What efficacy, safety, patient response, or clinical significance may follow?

The authority boundary is explicit:

P0–P2
    → direct Genome Physics scope after validation

P3
    → mechanistic interpretation requiring supporting evidence

P4–P5
    → higher biological and clinical layers requiring separate models and experiments

A local opening proxy is not complete accessibility.

A mutation delta is not pathogenicity.

A guide–target interaction is not editing outcome.

An oligo–target calculation is not in-vivo potency.

The Decision Object must name the claim level of every output.

10. The First Executable Scientific Contract

The first implementation target is deliberately bounded.

Given a short DNA or RNA system, its declared state, environment, and optional perturbation, calculate local geometry, pairing, stacking, mechanics, opening behavior, and physical deltas from Flux physics.

Required inputs

• sequence;
• DNA or RNA chemistry;
• strand and pairing state;
• modifications, if supported;
• temperature;
• salt and relevant ions;
• pH or protonation assumptions where required;
• solvent or hydration state;
• reference and altered state, when comparing a perturbation.

Required authoritative outputs

• base-pair geometry;
• dinucleotide-step geometry;
• twist and rise;
• roll and tilt;
• shift and slide where supported;
• bend and torsional response;
• pairing and stacking contributions;
• local opening or separation tendency;
• mismatch or mutation delta;
• state sensitivity;
• provenance and applicability.

Explicitly excluded from the first property claim

• pathogenicity;
• gene expression;
• complete cellular accessibility;
• chromatin state;
• editing efficiency;
• immune response;
• delivery;
• clinical effect.

Promotion rule

The first property layer may be promoted only after:

1. derivation and code are frozen;
2. state inputs are explicit;
3. no empirical property determinant enters the output;
4. retrospective physical benchmarks pass;
5. held-out sequence contexts pass;
6. external or blind evaluation passes;
7. limitations are published.

The detailed benchmark contract is provided in the accompanying Foundational Benchmark Blueprint.

11. Foundational Benchmark Architecture

The first benchmark program should be frozen before product implementation becomes adaptive.

It contains six packets.

Packet G — Geometry

Tests:

• base-pair geometry;
• dinucleotide-step geometry;
• twist, rise, roll, tilt, shift, and slide;
• local bend and width.

Primary evidence:

• high-resolution structures;
• declared state and environment;
• sequence-context holdouts.

Packet T — Thermodynamic and energetic behavior

Tests:

• pairing and stacking contributions;
• duplex or hairpin stability under declared conditions;
• ordering of sequence and mismatch effects.

Conventional nearest-neighbor models remain comparators only.

Packet M — Mismatch and mutation delta

Tests:

• single mismatches;
• short insertions and deletions where represented;
• sequence-context transfer;
• sign and rank of physical deltas.

Packet K — Mechanics

Tests:

• bending;
• torsional response;
• stiffness or persistence-related quantities;
• force or confinement response where data exist.

Packet E — Environment

Tests:

• temperature;
• salt and ion changes;
• pH or protonation changes where supported;
• environmental sensitivity without fitted correction formulas.

Packet R — Short RNA states

Tests:

• short stems and hairpins;
• mismatches and bulges;
• local RNA geometry;
• state ranking.

This packet may be a second-gate benchmark if the first release is DNA-duplex focused.

Split rules

Hold out by:

• sequence context;
• motif family;
• length;
• mismatch class;
• chemical modification;
• environmental condition;
• source laboratory where practical.

Near-duplicate sequences and structures must not cross development and test splits.

Metrics

Use property-appropriate metrics, including:

• MAE or RMSE in physical units;
• angular error;
• relative error;
• rank correlation;
• sign accuracy for deltas;
• top-k enrichment for candidate ranking;
• interval coverage for physically propagated uncertainty;
• failure-rate and applicability reporting.

Pass/fail thresholds must be fixed before hidden evaluation and tied to:

• experimental repeatability;
• accepted non-ML physical baselines;
• modern ML baselines;
• null sequence-composition baselines.

No baseline may be used to tune the final Flux property.

12. The Eight Planned Modules

The module family contains exactly eight planned interfaces.

Foundational kernels and research programs support the eight planned interfaces.

1. Genome Physics Scanner

Planned — first release

Intended output:

• position-by-position physical tracks;
• local structure and mechanics;
• mutation-sensitive regions;
• state and applicability warnings.

The Scanner is not a pathogenicity or expression engine.

2. Variant Mechanism

Planned — first release

Intended output:

• reference-versus-altered physical delta;
• changed structure or interaction;
• claim-level label;
• decisive validation experiment.

A physical delta is not automatically a clinical interpretation.

3. Oligo Triage

Planned — first release

Intended output:

• primary and backup ASO, siRNA, guide, primer, or probe families;
• target interaction;
• self-structure;
• mismatch discrimination;
• off-target physical panel;
• synthesis shortlist.

4. Oligo Designer

Planned — second release

Intended output:

• constrained redesigns;
• improved physical trade-offs;
• preserved chemistry and motif constraints;
• prospective assay plan.

5. Resistance Rescue

Planned — first release cross-interface workflow

Intended output:

• reference-versus-mutant target delta;
• retained and lost interactions;
• compound or sequence ranking;
• rescue hypotheses;
• frozen experimental panel.

6. Construct Optimizer

Planned — second release

Intended output:

• physical-risk map;
• self-binding and structural hotspots;
• preservation-aware redesign;
• manufacturing or assay hypothesis, separately labeled.

7. Protein–Nucleic-Acid Physics

Planned — later release

Intended output:

• mode-specific protein–DNA or protein–RNA interaction;
• mutation and binding-state deltas;
• validated structure and partner requirements.

8. Regulatory Designer

Planned — later release

Intended output:

• physical presentation and local structural state;
• protein-binding hypotheses;
• preservation-aware regulatory variants;
• reporter-validation plan.

It must not present physical presentation as complete expression prediction.

13. Foundational Capabilities

Several capabilities are necessary foundations but should not inflate the module count.

Nucleotide and Duplex Core

Required for:

• pairing;
• stacking;
• geometry;
• mechanics;
• mismatch;
• environment;
• modifications.

This is the physical kernel beneath the interfaces.

Structural-State Search

Required for:

• hairpins;
• stems;
• loops;
• self-dimers;
• alternative alignments;
• local DNA and RNA states.

External algorithms may propose states.

Flux physics must evaluate promoted states.

RNA Physics Core

Required for:

• short RNA structure;
• oligo self-structure;
• target opening;
• structured RNA;
• later aptamer and interaction work.

RNA Physics is a foundational capability feeding multiple modules, not necessarily a separate product card.

Environmental and Modification Layer

Required for:

• salt and ions;
• pH;
• temperature;
• hydration;
• backbone and base modifications.

These conditions define the physical state.

They may not become empirical tuning knobs.

14. Genome Physics Scanner

The Scanner is the intended first general interface.

It is planned to accept:

• DNA or RNA sequence;
• reference and variant data;
• construct sequence;
• declared environment;
• modification map.

It is planned to return position-by-position physical tracks such as:

• pairing and stacking state;
• local geometry;
• twist and bend;
• torsional strain;
• opening tendency;
• mismatch or mutation sensitivity;
• alternative-state risk;
• confidence and applicability.

Scanner boundaries

The first Scanner will not claim:

• gene expression;
• pathogenicity;
• complete chromatin state;
• whole-cell behavior;
• complete protein accessibility;
• editing outcome.

Use language such as:

• local physical opening;
• structural-risk proxy;
• physical presentation;
• mutation-sensitive region;
• biological hypothesis.

Long-sequence execution

The local kernel is intended to be:

• parallel;
• cacheable;
• deterministic;
• suitable for sliding-window analysis.

Whole-genome throughput remains a performance target until measured.

Do not publish unmeasured “millions per second” or whole-genome runtime claims.

15. Mutation Physics and Variant Mechanism

A mutation changes the represented physical system.

Relevant perturbations include:

• base substitution;
• mismatch;
• insertion;
• deletion;
• lesion;
• chemical modification;
• repeat expansion;
• protein mutation in cross-interface programs.

The primary output is a physical delta.

Reference state
    ↓ compare
Altered state
    ↓
Change in geometry, energy, mechanics, structure, or interaction

A Variant Mechanism result may contain:

• changed pairing or stacking;
• local structure delta;
• opening or accessibility proxy;
• interaction loss or gain;
• affected region;
• alternative state;
• claim level;
• confidence and missing context;
• proposed validation experiment.

The output must not collapse into a binary pathogenicity label.

The first question is:

What changed physically?

Only then may a biological hypothesis be proposed.

16. Oligo and Guide Triage

The first commercial sequence-design wedge is intended to reduce candidate spaces before synthesis.

Candidate classes include:

• antisense oligonucleotides;
• siRNA;
• CRISPR guides;
• primers;
• probes.

A triage program may consider:

• target interaction;
• duplex state;
• mismatch discrimination;
• self-hairpin risk;
• self-dimer risk;
• local target opening;
• off-target physical compatibility;
• mutation robustness;
• chemistry and length constraints;
• manufacturability.

Candidate generation

AI, enumeration, motif rules, and catalogs may propose candidates.

They may not determine the authoritative physical property.

Biological boundary

Oligo triage does not by itself establish:

• delivery;
• immune response;
• tissue uptake;
• intracellular trafficking;
• nuclease activity;
• complete in-vivo potency;
• patient response.

Those require Pharmacology, Materials, biological, and experimental layers.

Planned Decision Object

The output should contain:

• primary candidates;
• backup families;
• property breakdown;
• rejected candidates and reasons;
• mutation sensitivity;
• off-target panel;
• claim levels;
• proposed synthesis and assay panel.

17. Oligo Redesign

Once ranking survives prospective validation, the platform may move from analysis to constrained redesign.

A redesign program may seek to:

• improve target interaction;
• reduce self-structure;
• improve mismatch discrimination;
• reduce off-target compatibility;
• preserve required motifs or chemistry;
• improve mutation robustness;
• maintain manufacturability.

Existing candidate
    ↓
Physical failure analysis
    ↓
Constrained variants
    ↓
Authoritative Flux calculation
    ↓
Pareto-ranked redesigns
    ↓
Prospective assay

The redesign engine must separate:

• generated sequence;
• calculated property;
• decision score;
• biological hypothesis;
• assay result.

A generated sequence is not improved merely because the search algorithm proposed it.

It must be physically recalculated and prospectively tested.

18. Resistance Rescue

Resistance Rescue is the first major cross-interface Genome Physics workflow.

It combines:

• reference and mutated target state;
• Pharmacology interaction physics;
• Chemistry candidate properties;
• mutation representation;
• candidate search;
• assay design.

A program may ask:

Which compounds or sequence therapeutics retain interaction across this mutation panel, and what redesign may restore the lost mechanism?

The Decision Object should return:

• reference-versus-mutant target delta;
• retained and lost interactions;
• candidate ranking;
• primary and backup families;
• rescue hypotheses;
• uncertainty and state assumptions;
• frozen assay panel.

Every property must retain its interface provenance.

A mutation description from Genome Physics does not erase the Pharmacology evidence requirements of the interaction result.

19. Therapeutic Construct Architecture

Therapeutic constructs may contain physical risks that appear before biological testing or manufacturing.

Relevant systems include:

• mRNA;
• plasmids;
• viral-vector payloads;
• DNA vaccines;
• synthetic cassettes;
• promoter–coding–UTR constructs;
• repeated or structured regions.

A construct program may search for:

• local self-binding;
• stems and hairpins;
• unusually stable or unstable regions;
• bend and torsion hotspots;
• repeat-driven alternative states;
• mutation-sensitive segments;
• structure-associated manufacturing risks;
• redesigns preserving the intended function.

Preservation contract

Every redesign must declare what may not change:

• amino-acid sequence;
• guide function;
• regulatory motif;
• restriction site;
• manufacturing constraint;
• vector architecture;
• modification policy.

Manufacturing and expression boundary

A physical-risk calculation may support a manufacturing or expression hypothesis.

It does not establish:

• yield;
• expression;
• shelf life;
• delivery;
• immunogenicity;
• clinical performance;

without prospective evidence.

20. Protein–Nucleic-Acid Physics

Many high-value biological decisions depend on protein–nucleic-acid interaction.

Potential systems include:

• transcription factors;
• nucleases;
• polymerases;
• ribosomes;
• RNA-binding proteins;
• DNA-repair machinery;
• protein–oligo complexes.

Each interaction requires:

• explicit sequence state;
• explicit protein state;
• binding mode;
• environment;
• mode-specific Flux-derived closure;
• independent evidence.

A target name is not a target state.

Higher-authority calculations may require:

• experimental structure;
• validated model;
• mutation;
• conformation;
• cofactor;
• complex state.

This module remains a later release because it depends on validated nucleic-acid and protein-interaction foundations.

21. Regulatory Sequence Physics

Regulatory sequences present physical states to molecular machinery.

Potential targets include:

• promoters;
• enhancers;
• UTRs;
• splice-adjacent regions;
• terminators;
• ribosome-binding regions;
• compact synthetic regulatory modules.

A first physical layer may calculate:

• local structure;
• opening tendency;
• motif presentation;
• bend and torsion;
• competing states;
• protein-binding presentation;
• condition sensitivity.

Expression boundary

A physical accessibility or presentation result is not automatically expression prediction.

Expression may depend on:

• transcription factors;
• chromatin;
• polymerase;
• epigenetic state;
• cell type;
• RNA stability;
• transport;
• feedback and regulation.

Regulatory design should advance in layers:

1. physical structure and presentation;
2. protein–nucleic-acid interaction;
3. reporter or binding validation;
4. condition-specific biological models;
5. broader design only after prospective success.

22. Longer-Term Research Programs

Two longer-term programs remain outside the eight planned modules.

Chromatin and multiscale genome context

Potential layers include:

• nucleosomes;
• chromatin fiber;
• supercoiling;
• loops;
• topological domains;
• long-range contacts;
• chromosome mechanics;
• nuclear environment.

The platform should not begin with an atomistic chromosome.

It should build a hierarchy:

Local nucleotide physics
    ↓
Short structural states
    ↓
Protein and nucleosome interaction
    ↓
Coarse-grained chromatin
    ↓
Loops and domains
    ↓
Long-range genome context

Every coarse-grained representation must declare what it retains, discards, and inherits.

Chromatin and chromosome outputs remain a research frontier until lower layers validate.

Genome Compiler

The long-term inverse-design program translates a bounded biological or therapeutic objective into physically constrained sequence design.

Example requests may include:

• propose oligos robust to a mutation panel;
• reduce self-structure while preserving encoded protein;
• design a guide family with improved mismatch discrimination;
• preserve required motifs while reducing structural risk.

The Compiler should return:

• candidate sequences;
• physical-property traces;
• preserved constraints;
• trade-offs;
• rejected designs and reasons;
• claim levels;
• proposed experiments.

The Genome Compiler is a bounded research program—not a current product and not an autonomous organism-design system.

23. The Genome Physics Decision Object

A Genome Physics Decision Object is the primary output of a sequence or interaction workflow.

It may contain:

• normalized sequence and modification state;
• strand, structure, environment, and partner assumptions;
• Foundation Interface version;
• Genome Physics Domain Closure versions;
• authoritative physical properties;
• generated structural candidates;
• inferred states;
• mutation or design deltas;
• candidate rankings;
• rejected candidates and reasons;
• claim-level labels;
• biological hypotheses, separately identified;
• confidence and applicability;
• readiness status;
• evidence links;
• recommended experiment.

The object must separate six layers.

Authoritative physical property

Produced by the approved Flux calculation.

Generated structural or sequence candidate

Proposed by enumeration, AI, external folding, or search.

Inferred state

Pairing, structure, environment, or partner state selected by the Compiler.

Decision output

Candidate ranking, redesign recommendation, risk class, or assay priority.

Biological hypothesis

Interpretation beyond the directly calculated physical layer.

External evidence

Structure, assay, annotation, benchmark, or external model result with separate provenance.

This separation prevents a sequence workflow from collapsing into one unexplained score.

24. The Genome Physics Trust Layer

Genome Physics must be auditable from sequence input to experimental decision.

The Trust Layer includes:

Property Evidence Ledger

Records which sources may determine each property.

Formula and closure status

Classifies relations as:

• production Flux derivation;
• validated Flux-derived Genome Physics Domain Closure;
• planned;
• research hypothesis;
• non-authoritative aid;
• external comparator;
• validation pending;
• non-compliant;
• retired.

State-provenance record

Declares:

• DNA or RNA chemistry;
• sequence and modifications;
• strand and pairing state;
• structural state supplied or inferred;
• environment;
• interaction partner;
• mutation or redesign;
• assay context.

Claim-level register

Labels each output P0 through P5.

Readiness registry

Separates:

• foundation available;
• planned first release;
• planned second release;
• planned later release;
• research frontier.

Benchmark manifest

Defines:

• cohort;
• exclusions;
• state policy;
• hidden split;
• metric;
• comparator;
• pass/fail threshold.

Gap registry

Records missing physics and biological context.

Version record

Freezes the complete scientific and computational state used to produce a result.

A module does not advance because its interface looks complete.

It advances because a frozen claim survived the required evidence gate.

25. Readiness Architecture

Flux Genome Physics is currently at the architecture and foundational-benchmark-contract stage.

No module is commercially available today.

Foundation available in the adjacent platform

Reusable infrastructure includes:

• molecular and atomic representations;
• Chemistry calculations;
• protein and docking workflows;
• Pharmacology decision infrastructure;
• mutation and candidate enumeration;
• high-throughput execution;
• provenance and validation systems;
• APIs and reporting.

These are dependencies.

They are not Genome Physics products.

Planned — first release

• Genome Physics Scanner;
• Variant Mechanism;
• Oligo Triage;
• Resistance Rescue;
• Nucleotide and Duplex Core.

Planned — second release

• Oligo Designer;
• Construct Optimizer;
• expanded RNA and structural-state foundation.

Planned — later release

• Protein–Nucleic-Acid Physics;
• Regulatory Designer.

Research frontier

• chromatin and chromosome-scale physics;
• long-range regulatory context;
• Genome Compiler.

Advancement rule

A planned capability advances only after:

1. property evidence is derived;
2. implementation and state schema are frozen;
3. retrospective physical evidence passes;
4. held-out or blind evidence passes;
5. prospective evidence passes for ranking or design claims;
6. the commercial decision improves measurably;
7. safety and biosecurity review passes.

26. Validation Strategy

Genome Physics must be built through evidence gates.

Gate 1 — Mathematical consistency

Does the implementation reproduce the governing derivation, units, and state contract?

Gate 2 — Retrospective physical benchmark

Does it reproduce structural, energetic, mechanical, or interaction measurements not used to determine the property?

Gate 3 — Held-out sequence context

Does it generalize across withheld:

• sequence contexts;
• motif families;
• lengths;
• mismatches;
• modifications;
• environmental conditions?

Gate 4 — Blind external validation

Can an external group choose hidden cases and independently score frozen predictions?

Gate 5 — Prospective candidate ranking

Are top-ranked oligos, guides, compounds, or constructs enriched for measured success?

Gate 6 — Prospective redesign

Does a generated redesign improve a predefined endpoint?

Gate 7 — Operational value

Does the platform reduce:

• candidates;
• synthesis;
• assays;
• design cycles;
• time;
• uncertainty;
• program cost?

Failure handling

A failed prediction is not repaired by fitting the benchmark.

It triggers:

• implementation repair;
• derivation repair;
• missing-physics work;
• state-model improvement;
• or claim restriction.

Independent Validation

Suitable first challenges include:

• hidden duplex geometry;
• mismatch energetics;
• local opening or melting behavior under declared conditions;
• short RNA state ranking;
• oligo ranking;
• guide mismatch discrimination;
• resistance mutation panels;
• construct-risk assays.

Independent Validation

27. Development Sequence

The sections below set the scientific dependency order.

Development follows the workstreams, evidence gates, and safety controls defined by The Genome Physics Initiative.

Stage A — Freeze the scientific contract

• state schema;
• property definitions;
• authority rules;
• benchmark packets;
• hidden splits;
• pass/fail thresholds;
• disclosure and safety boundaries.

Stage B — Build the physical foundation

• nucleotide representation;
• duplex state;
• pairing and stacking;
• geometry and mechanics;
• mismatch and environment;
• foundational validator.

Stage C — Build the first decision surfaces

• Scanner;
• Variant Mechanism;
• Oligo Triage;
• Resistance Rescue.

Stage D — Move from ranking to redesign

• Oligo Designer;
• Construct Optimizer;
• prospective design evidence.

Stage E — Add complex interaction and regulation

• Protein–Nucleic-Acid Physics;
• Regulatory Designer;
• expanded RNA and structural search.

Stage F — Enter research-frontier multiscale design

• chromatin;
• long-range genome context;
• Genome Compiler.

The detailed roadmap is maintained separately:

The Genome Physics Initiative

28. First Commercial Problems

The first customer problems must remain bounded and experimentally measurable.

Resistance mutation impact and rescue

Question

Which compounds or sequence therapeutics retain interaction across a mutation panel, and what redesign may restore lost interaction?

Planned output

• reference-versus-mutant target delta;
• retained and lost interactions;
• candidate ranking;
• rescue hypotheses;
• frozen assay panel.

ASO, siRNA, and guide triage

Question

Which small set of candidates should be synthesized first?

Planned output

• target interaction;
• self-structure;
• mismatch discrimination;
• off-target physical compatibility;
• mutation robustness;
• primary and backup families.

Therapeutic construct risk scanning

Question

Which regions of a construct create physical risk, and which redesigns preserve the required function?

Planned output

• structural-risk map;
• self-binding;
• unstable or over-stabilized regions;
• bend and torsion hotspots;
• preservation-aware redesigns.

Commercial value

The product is reduced uncertainty.

Value may come from:

• fewer candidates synthesized;
• fewer assays;
• earlier rejection;
• mechanism-level failure explanation;
• faster redesign;
• better backup candidates;
• fewer manufacturing surprises;
• improved prospective enrichment.

The first commercial claim is not “predict all biology.”

It is:

Reduce the physical design space before expensive experiments begin.

29. What Success Looks Like

The first proof is not a patient prediction or a whole-genome model.

It is a sequence of increasingly strong results.

Physical proof

A frozen kernel predicts unseen geometry, state, mismatch, mechanics, or environmental response.

Ranking proof

Top-ranked candidates are enriched for measured success relative to baseline selection.

Mechanism proof

The platform identifies why a mutation, oligo, guide, or construct fails.

Redesign proof

A generated candidate improves a predefined experimental endpoint.

Workflow proof

A partner reduces synthesis, assays, or design cycles.

Illustrative evidence goals include:

• a large candidate set reduced to a small testable panel;
• top-ranked oligos enriched for activity;
• resistance mutations separated by candidate impact;
• a redesigned sequence outperforming its starting candidate;
• a construct redesign improving a declared stability or manufacturing measure;
• a predicted physical mechanism confirmed by a decisive experiment.

These are success definitions.

They are not current achievements.

30. Claims Require Evidence

Flux Genome Physics will not be presented as:

• a clinical diagnostic;
• a universal pathogenicity predictor;
• a guarantee of in-vivo efficacy;
• a delivery or immune-safety solution;
• complete expression prediction;
• complete editing-outcome prediction;
• whole-cell prediction from sequence;
• autonomous organism design;
• unrestricted genome optimization;
• a replacement for laboratory validation.

It will not claim:

• whole-genome authority from a local kernel;
• regulatory authority from an opening proxy;
• clinical effect from a mutation delta;
• editing efficiency from duplex physics alone;
• therapeutic potency from target interaction alone.

Higher-layer claims inherit the uncertainty and evidence status of every layer beneath them.

31. AI, Sequence Models, and External Tools

Sequence AI can learn powerful patterns from:

• biological sequence;
• structure;
• assays;
• expression;
• variants;
• literature.

Flux Genome Physics is intended to provide a complementary physical layer.

AI may

• generate candidate sequences;
• enumerate modifications;
• prioritize search;
• propose structures;
• organize literature;
• write and test code;
• choose which candidate receives a Flux calculation first;
• summarize Decision Objects.

Limits on AI's role

• determine the authoritative pairing, stacking, geometry, structure, interaction, or mutation property;
• repair a residual;
• substitute a learned fold energy;
• replace a physical off-target calculation;
• promote a candidate without authoritative recomputation.

External scientific tools may

• propose structural states;
• provide external comparisons;
• serve as baselines;
• support cross-checking;
• identify candidates for higher-detail analysis.

They retain separate provenance.

The division of labor is:

AI proposes. Matter Computing computes. Experiments decide.

32. Safety and Biosecurity Boundary

Genome Physics is intended for legitimate scientific, therapeutic, diagnostic-research, and engineering use.

Its architecture must be bounded before inverse sequence design becomes operational.

Supported use classes

Examples include:

• therapeutic oligo design;
• mutation and resistance analysis;
• research-use guide and probe triage;
• construct stability and manufacturability research;
• bounded regulatory and reporter studies;
• biosensor and assay design;
• external validation of physical predictions.

Escalated or prohibited use classes

The platform must not provide unrestricted assistance for:

• pathogen optimization;
• virulence enhancement;
• host-range expansion;
• immune evasion;
• harmful payload design;
• autonomous whole-organism design;
• unsafe whole-genome synthesis;
• other high-risk biological engineering.

Governance requirements

Sequence-design workflows should include:

• stated scientific purpose;
• user and organization review where appropriate;
• target and organism context;
• bounded design scope;
• audit logging;
• human review before synthesis or experimental handoff;
• escalation for high-risk requests;
• refusal or restriction where safety requirements are not met.

Output and safety boundary

This account covers the platform’s principles and planned capabilities. It does not expose protected Flux Theory mechanisms, and the platform does not provide unrestricted high-risk sequence designs.

Safety is not an appendix added after the Genome Compiler exists.

It is part of the Compiler contract.

33. Genome Physics as Platform Infrastructure

Flux Genome Physics will strengthen every adjacent domain.

It extends Chemistry

• nucleotide chemistry;
• pairing and stacking;
• backbone mechanics;
• sequence-dependent environment;
• ligand–nucleic-acid interaction.

It extends Pharmacology

• resistance mutations;
• oligo therapeutics;
• target opening;
• protein–nucleic-acid interaction;
• regulatory targets;
• therapeutic sequence design.

It extends Materials

• delivery systems;
• biosensors;
• nanoparticle and surface interactions;
• physical assay environments;
• biological–material interfaces.

It extends the Matter Compiler

• sequence as a programmable physical object;
• mutation as a graph transform;
• preservation-aware redesign;
• claim-level governance;
• multiscale biological state;
• new inverse-design workflows.

Genome Physics is not the endpoint of Matter Computing.

It is the layer that begins connecting matter to biological design.

34. Current Genome Physics Priorities

Flux Genome Physics should advance through nine priorities.

1. Derive the local kernel

Establish pairing, stacking, geometry, mechanics, mismatch, and environment from Flux physics.

2. Keep sequence state explicit

Do not let minimal input hide strand, structure, modification, environment, or interaction assumptions.

3. Freeze the benchmark before optimizing

Define cohorts, splits, metrics, and thresholds before hidden evaluation.

4. Build local before global

Start with bounded windows and short structures before chromatin or chromosome claims.

5. Separate physics from biological inference

Keep P0–P2 outputs distinct from P3–P5 hypotheses and outcomes.

6. Preserve authority through search

Allow AI and external tools to propose states and candidates, but require authoritative Flux recomputation.

7. Validate before redesign

Prove unseen prediction and ranking before inverse design claims.

8. Freeze prospective predictions

Use hidden assays and partner-held targets to test ranking and redesign.

9. Build safety into the Compiler

Bound purpose, target, output, and experimental handoff from the beginning.

Progress should be measured by:

• how many nucleic-acid properties emerge from the shared foundation;
• how many sequence-state assumptions become explicit;
• how many planned modules pass blind validation;
• how many candidate spaces become searchable;
• how many prospective designs succeed;
• how much uncertainty and experimental burden are reduced;
• how much Genome Physics strengthens Pharmacology and therapeutic design.

35. Conclusion

Sequence is not only information.

It is matter in a state.

That state determines:

• structure;
• mechanics;
• interaction;
• mutation response;
• design possibility.

Flux Genome Physics is not another sequence model, genomic score, folding database, or disconnected therapeutic-design application.

It is the planned DNA, RNA, mutation, and therapeutic-sequence interface to one Matter Computing Platform.

Its scientific standard is universal:

Every authoritative physical property must emerge from Flux physics.

Its state discipline is explicit:

Sequence identifies the matter. State determines the physical problem.

Its claim discipline is layered:

Physics before phenotype. Mechanism before outcome.

Its development discipline is strict:

Predict before optimizing. Local before global.

Its operational purpose is practical:

Reduce the physical design space before expensive experiments begin.

Its strategic role is the next frontier:

Genome Physics extends Matter Computing from molecular and therapeutic interaction into biological sequence and design.

Flux Theory opened the door.

Matter Computing made the next frontier visible.

FluxMateria is building it.

Appendix A — Genome Physics Terminology

Appendix B — Property Evidence Matrix

Appendix C — Module Readiness, July 2026

No commercially available Genome Physics module exists today.

Appendix D — Claim-Level Matrix

Appendix E — Validation Gates

A capability may not skip a gate merely because its interface appears complete.