mokume

Reference benchmark datasets for proteomics DEA optimization. Contains ground-truth metadata, expected fold-changes, and best-known method configurations. Used by the evaluator to score candidate workflows against known references.

Domain heuristics for proteomics differential expression analysis. Covers data-type priors (LFQ, TMT, DIA), setting-specific top methods, ensemble strategies, and conditional rules derived from a 34576-experiment benchmark. Used by proposer and reflector to narrow the search space.

LLM prompt templates for proposing proteomics DEA configurations. Provides the system and user prompts used by the proposer to generate candidate analysis pipelines via tool calling.

LLM prompt templates for reflecting on proteomics DEA experiment results. Guides the reflector to analyze score trajectories, identify parameter drivers, declare convergence, and propose targeted perturbations.

LLM prompt templates for generating final analysis reports. Produces a concise Markdown report covering executive summary, recommended configuration rationale, key findings, and caveats.