Use this runbook to turn a pathway and species question into a comparative evidence packet.
Record the target pathway, target species, comparison species, expected decision, exclusions, and review limit.
Check public genome, annotation, transcriptome, protein, and literature sources. Record stable identifiers, versions, access dates, and acquisition rules.
Stop if the available data cannot support the requested claim. Select a weaker route or design the next experiment.
Build the query ledger from characterized source proteins or reviewed public sequences. Add positive, negative, and broad-family controls.
Do not use unresolved names as sequence identifiers.
Choose annotation-direct, genome-context, transcript-first, transcriptome-only, synteny, rescue, or next-experiment design.
Record rejected routes and the claim limit.
Run bounded sequence, profile, structure, or domain searches. Keep each evidence channel separate. Record commands, versions, parameters, inputs, outputs, and hashes.
Add genome coordinates, neighborhoods, synteny, expression, or BGC calls only when the source data supports them.
Transcript evidence can nominate candidates. It cannot prove physical clustering.
Normalize identifiers before joins. Keep caller disagreements and missing values. Do not average incompatible scores into one unexplained rank.
Include candidate tables, cluster views, pathway coverage, provenance, conflicts, claim limits, and next actions.
Run the declared checks. Record the outcome, artifacts, versions, hashes, limits, and next bounded action.