Compare expression between groups from a reviewed count matrix

Why use this

1. A researcher comparing experimental groups needs sample labels and statistical choices to remain traceable from the count matrix to the eventual figures. A mislabeled sample or poorly specified contrast can undermine an otherwise orderly analysis package.

2. This job could prepare a reproducible analysis specification and preflight review before any statistics are run. Making formula, contrast and metadata assumptions explicit helps experts decide whether the planned comparison addresses the scientific question.

Why an agent / simpler alternative

Useful judgment

Reconcile sample metadata, expose ambiguous contrasts and organize approved analysis choices without substituting for statistical judgment.

Simpler alternative

Use a validated lab workflow when the design and count inputs match its documented assumptions.

Unsuitable when

The comparison is not scientifically justified or protected biological data cannot be processed within approved boundaries.

Proposed inputs

1. Name

   Count/metadata manifest

   Owner

   Authorized researcher

   Minimum scope

   Matrix dimensions, sample IDs and raw-count provenance

   Quality check

   Check identifier alignment and duplicate samples

2. Name

   Analysis design

   Owner

   Qualified lead

   Minimum scope

   Chosen formula, contrast and relevant covariates

   Quality check

   Confirm reference levels and design meaning

Access and setup — proposed or unknown

Proposed minimum

1. Review a synthetic manifest and proposed formula/contrast; no statistical execution, installation or output overwrite.

Compatibility unknowns

1. Current analysis-library versions, data support, compute resources and runtime compatibility are unknown.

External actions requiring approval

1. Data uploads, computational spending, analysis changes and replacing output directories require authorization.

Expected deliverable and review checks

Proposed artifact

Analysis-specification candidate with preflight exceptions, parameter manifest and expected result/replay artifacts.

Critical checks

1. Align samples exactly; verify contrast direction and covariate definitions; retain unresolved design issues rather than producing nominal results.

Missing output evidence

This overlay adds no new output inspection, execution or validation; consult the referenced canonical source outcome limits.

Failure signals and proposed responses

1. Signal

   Metadata rows are aligned by position instead of ID

   Why it matters

   Groups may be assigned to the wrong samples

   Proposed response

   Require explicit identifier matching.

2. Signal

   Condition and another factor cannot be separated in the design

   Why it matters

   The intended comparison may be uninterpretable

   Proposed response

   Escalate the design question to a qualified statistician.

Effort and cost — unknown

Setup work

1. Document provenance and the approved design.

Ongoing review

1. Experts review quality diagnostics and statistical outputs from any future authorized run.

Unknowns

1. Design adequacy, data quality, computational needs and analysis costs are unknown.

First check — proposed, unexecuted

Status

proposed_unexecuted

Synthetic fixture

Invent a count manifest with shuffled sample IDs and a contrast whose reference label is missing; perform a paper preflight.

Acceptance checks

1. Flag alignment/reference problems; run no model; infer no expression differences.

Stop conditions

1. Stop if sample identity or scientific design remains unresolved.

Limits

1. A preflight specification cannot establish statistical validity, biological effects or reproducibility of unexecuted analyses.