bijux-proteomics-lab

bijux-proteomics-lab converts a supported recommendation into an operational assay plan and records what the experiment actually observed. It owns assay design, readiness, batching, scheduling, handoff, outcome capture, and feedback to scientific review.

python -m pip install bijux-proteomics-lab

Authority before material consequence

An analytical recommendation is input to laboratory planning, not permission to consume material or operator time. Each authority can stop progression for a different reason.

Authority	Decision	Required record	What it cannot override
Intelligence reviewer	is the proposed action defensible enough to consider?	recommendation, alternatives, conditions, uncertainty, and human-review state	assay answerability or operational capacity
scientific assay owner	can the proposed measurement answer the stated question?	assay design, controls, acceptance criteria, confounders, and refusal checks	material custody, safety, or scheduling authority
operational owner	can the work be executed with identified people, material, equipment, capacity, and timing?	readiness, resource, batch, schedule, and assigned-owner records	scientific sufficiency or required safety approval
human or safety authority	is action permitted under the applicable risk and governance boundary?	approval, hold, escalation, or refusal with accountable owner	missing controls or an uninterpretable design
measurement reviewer	did the observation satisfy QC and acceptance criteria?	raw and processed observations, QC, deviations, missingness, and disposition	the meaning of the result for the upstream claim
Knowledge and Intelligence review	how should accepted consequence evidence affect claims and future action?	appended evidence, reconciliation, revised review, and policy-bound decision	historical plan, recommendation, or observation records

Progression requires every applicable authority; one approval cannot compensate for another authority's refusal. Holds and refusals remain first-class outputs with a named closure condition.

Closed-loop validation

flowchart LR
    recommendation["recommendation record"]
    design["assay design\nprotocol · controls · materials"]
    readiness{"ready?"}
    authorize{"human and operational authority?"}
    queue["priority · batches · schedule"]
    handoff["executable lab handoff"]
    observation["observed outcome"]
    acceptance{"measurement and controls accepted?"}
    reconcile["requested vs observed\nreconciliation"]
    feedback["new evidence and policy feedback"]
    refusal["refusal · hold · revised plan"]
    rejected["rejected or failed observation record"]

    recommendation --> design --> readiness
    readiness -->|yes| authorize
    readiness -->|no| refusal
    authorize -->|yes| queue --> handoff --> observation --> acceptance
    authorize -->|no| refusal
    acceptance -->|yes| reconcile --> feedback
    acceptance -->|no| rejected --> feedback

Planning and observation are separate records. A recommendation can be scientifically plausible but operationally unready; an executed assay can be technically successful but inconclusive; and an unexpected outcome can weaken the upstream evidence or decision policy.

Custody chain

An executable handoff crosses from analytical advice into material and operational consequence. Custody therefore remains explicit at every stage:

Stage	Accountable record	Required evidence before progression
requested	recommendation and scientific question	target, rationale, uncertainty, requested measurement
designed	assay plan	protocol, controls, materials, acceptance criteria, known risks
ready	readiness decision	answerability, completeness, capacity, ownership, refusal checks
scheduled	batch and schedule	compatibility, priority, resources, timing, assigned operator
handed off	immutable handoff record	exact instructions, identities, custody acknowledgement
observed	observation record	raw and processed measurements, QC, deviations, failures
reconciled	consequence record	requested-versus-observed comparison and disposition

A later record references its parents. It does not rewrite a recommendation, plan, or observation to make the history appear more certain.

Operational capabilities

Surface	Responsibility
design	experiment and protocol contracts
planning	advisory and executable assay plans, priorities, queues, batches, schedules, and next-cycle planning
readiness	stage-specific operational checks and progression decisions
lifecycle	governed movement through lab states
handoffs	artifacts, explanations, serialization, PTM and QC feedback, risk, and transfer records
outcomes	observations and evidence feedback
reconciliation	requested-versus-observed follow-up and flagship closure
benchmarks	lab claims, rehearsals, follow-up evidence, learning, and outcome dossiers

The package root exposes three planning entry points:

from bijux_proteomics_lab import (
    build_advisory_assay_plan,
    build_executable_assay_plan,
    plan_experiment_batches,
)

An advisory plan can be produced while operational details remain incomplete. An executable plan requires the stronger readiness and handoff information needed to act. Batch planning groups ready work under declared constraints; it does not make an unready assay executable.

Plan and record types

Artifact	Meaning	Authority
advisory assay plan	scientifically relevant follow-up with unresolved operational detail	supports review and refinement only
executable assay plan	protocol, controls, materials, acceptance, and ownership satisfy readiness policy	eligible for scheduling and authorized handoff
batch plan	ready assays grouped under capacity and compatibility constraints	allocates work; does not change assay readiness
handoff record	exact instructions, identities, materials, risks, and acceptance criteria transferred to an operator	binds the planned work to execution custody
observation record	what was measured, including QC, deviations, missingness, and failures	records fact without interpreting upstream consequence
reconciliation record	requested-versus-observed comparison and disposition	feeds new evidence into scientific review

Readiness and refusal

Readiness evaluates scientific inputs, controls, materials, protocol detail, capacity, scheduling constraints, and handoff completeness. A refusal is an expected scientific safety result when a plan would waste material, obscure interpretation, or create an outcome that cannot answer the stated question.

A refusal record identifies the blocking condition and a valid next action such as narrowing the assay, adding controls, resolving an upstream contradiction, rerunning analysis, or waiting for capacity. It is an operational and scientific disposition, not a generic runtime failure.

flowchart TD
    P["proposed assay"] --> S{"scientific question answerable?"}
    S -->|no| U["return upstream"]
    S -->|yes| C{"controls and acceptance complete?"}
    C -->|no| N["revise plan"]
    C -->|yes| M{"materials, capacity, and risk acceptable?"}
    M -->|no| Q["queue, narrow, or refuse"]
    M -->|yes| H["authorize executable handoff"]

Readiness is evaluated at the declared stage. Passing design review does not imply material availability, scheduling approval, execution success, or an interpretable outcome.

Outcome feedback

Observed outcomes are reconciled against the requested measurements and acceptance criteria. The resulting consequence record can:

• strengthen or weaken a knowledge claim;
• expose an unmodeled contradiction;
• recalibrate candidate ranking or recommendation confidence;
• trigger a revised assay plan;
• close a follow-up route as confirmed, rejected, or inconclusive.

History is append-only in meaning: later evidence can supersede a conclusion without erasing the recommendation and assumptions that led to the experiment.

Evidence promotion contract

An observation crosses three independent review boundaries before it can alter an action. Lab records the measurement and its assay disposition; Knowledge decides its relationship to a claim; Intelligence decides whether the revised evidence changes the recommendation.

Boundary	Record entering	Decision produced	Decision explicitly not produced
assay acceptance	raw and processed measurements, controls, QC, deviations	accepted, rejected, failed, or inconclusive observation	biological support or contradiction
claim reconciliation	accepted observation with subject, context, and lineage	support, contradiction, qualification, context, or gap	candidate rank or permission to act
decision revision	versioned evidence bundle and prior decision record	rerank, retain, downgrade, escalate, hold, or refuse	laboratory authorization

sequenceDiagram
    participant L as Lab
    participant K as Knowledge
    participant I as Intelligence
    participant H as Human authority
    L->>K: accepted observation and assay limits
    K->>I: versioned claim relationships
    I->>H: revised advisory disposition
    H-->>L: authorized plan, revised question, or stop

Rejected and inconclusive observations remain durable records, but they do not skip assay acceptance by being narratively useful. A new action always returns through review and authorization.

Interpret the outcome safely

Execution state, assay QC, answerability, and biological consequence are separate judgments. The reconciliation record retains all four.

Observed condition	Lab disposition	Downstream meaning
work did not start because readiness failed	refused or deferred	no observation; return the blocking condition to the plan owner
execution failed before a valid measurement	operational failure	no scientific conclusion; preserve diagnostics and consumed resources
measurements exist but controls or QC fail	observed, not accepted	retain data and deviations; do not promote to claim evidence
QC passes but the requested contrast is unresolved	technically accepted, inconclusive	record answerability limits and revise the question or assay
accepted observation supports the expected direction	consequence supports the requested proposition	Knowledge determines the resulting support edge and its scope
accepted observation opposes the expected direction	consequence contradicts or qualifies the proposition	preserve the unexpected result and trigger review
accepted observation falls outside the planned interpretation	unmodeled consequence	create a knowledge gap; do not force it into confirm or reject

flowchart TD
    OR["observation record"] --> QC{"measurement and controls accepted?"}
    QC -->|no| OA["observed, not scientifically accepted"]
    QC -->|yes| AN{"requested question answered?"}
    AN -->|no| IC["inconclusive consequence"]
    AN -->|yes| RC["requested-versus-observed reconciliation"]
    RC --> SP["support, contradiction, qualification, or gap"]
    SP --> KR["Knowledge review"]

The disposition must describe the observed record, not the hoped-for result. Unexpected and inconclusive outcomes remain first-class evidence about the assay, the question, or the upstream model.

Observation is not interpretation

Lab owns what was requested, executed, measured, and observed under the assay contract. Knowledge owns how the observation supports or contradicts a claim; Intelligence owns whether the new evidence changes a ranking or action. This separation prevents a technically clean measurement from being promoted directly into biological certainty.

sequenceDiagram
    participant L as Lab observation
    participant K as Knowledge review
    participant I as Intelligence judgment
    L->>K: append measurement, QC, deviations, lineage
    K->>K: reconcile support and contradiction
    K->>I: versioned evidence bundle
    I->>I: rerank, downgrade, or refuse

Learn From Non-Confirming Work

A closed loop does not require a positive biological result. A readiness refusal, execution failure, QC rejection, or inconclusive observation can still protect material, expose a weak question, and produce evidence for the next decision—provided its disposition is not collapsed into “no result.”

Disposition	What was learned	Safe next route
readiness refused	the plan cannot answer the question safely or completely under current conditions	add controls, resolve inputs, narrow scope, or stop
deferred	the plan may be valid but capacity, material, timing, or ownership is unavailable	retain priority and prerequisites without claiming execution
execution failed	the handoff began but did not yield an acceptable measurement	preserve diagnostics, consumed resources, and recovery boundary
observed, QC rejected	measurements exist but do not meet the assay contract	retain observations as rejected evidence and revise execution or controls
accepted, inconclusive	the assay worked but did not resolve the requested contrast	revise the question, power, measurement, or interpretation window
accepted, unexpected	the observation falls outside the anticipated support/refute frame	create a knowledge gap and review without forcing a binary label

flowchart TD
    attempt["planned follow-up"] --> disposition{"recorded disposition"}
    disposition -->|refused or deferred| planning["planning evidence"]
    disposition -->|failed| operational["operational evidence"]
    disposition -->|QC rejected| rejected["rejected observation"]
    disposition -->|inconclusive| question["question-design evidence"]
    disposition -->|unexpected| gap["new knowledge gap"]
    planning --> next["bounded next decision"]
    operational --> next
    rejected --> next
    question --> next
    gap --> next

The Workflow Refusal Handbook maps blocked work to safe responses, while Outcome Learning Loops returns accepted and non-confirming outcomes to evidence and policy review. Use Workflow Consequence Maps to connect those outcomes to the workflow family’s next decision burden.

Continue By Laboratory Question

Need	Read next	Review is complete when
evaluate cost, controls, and downstream burden	lab consequence	feasibility, control burden, cost of error, and allowed action share one consequence record
return outcomes to evidence and policy	outcome learning loops	accepted, rejected, inconclusive, and unexpected outcomes reach their correct owner without rewriting history
map a refusal to a safe next action	workflow refusal handbook	the blocker, protected resource, responsible owner, and admissible next route are explicit
separate planning, readiness, handoff, and outcomes	architecture	every stage has its own identity, parent record, authority, and terminal disposition
choose Python or artifact contracts	interfaces	the plan or outcome preserves controls, custody, deviations, QC, and refusal information
review operational and scientific bounds	known limitations	technical completion is not presented as answerability, biological support, or authority to act

Lab does not own upstream scientific computation, evidence truth, recommendation policy, or general execution orchestration.