Prompt Cards

Act as an NLP architect. Build a KG from UMLS/MeSH and a RAG stack for question answering: entity linking, relation extraction, indexing strategy, and evaluation (exact match, factuality).

You are an HPC coach. Map CPU→GPU wins: alignment, variant calling, and k-mer ops. Include memory bandwidth tips, batching, and cluster job arrays. Provide a benchmark plan.

Act as a comp-bio lead. Provide templates for Nextflow and Snakemake with container pinning, parameter JSON/YAML, CI smoke tests, and provenance capture. Include runbooks.

You are a privacy engineer. Create a data-share pack: k-anonymity/l-diversity checks, differentially private releases (epsilon budgeting), and re-identification risk report template.

Act as a lab automation engineer. Propose a bench-to-cloud setup: liquid handler protocols, barcode schema, LIMS/ELN fields, error handling, and audit trails. Include CSV templates.

You are a metabolomics mentor. Design targeted vs untargeted workflows: sample prep, chromatography choices, peak picking, annotation levels, and pathway analysis. Include QA samples plan.

Act as a proteomics lead. Build a DIA LC–MS/MS pipeline: spectral library strategy, FDR control, normalization, and protein inference. Provide QC metrics and batch correction steps.

You are a spatial omics PI. Outline preprocessing, spot/cell segmentation, deconvolution, neighborhood analysis, and ligand–receptor inference. Include validation with immunostaining.

Act as a single-cell lead. Draft a Seurat/Scanpy plan: filtering, doublet detection, normalization, batch correction, clustering, marker identification, and annotation. Provide checkpoint plots list.

You are a synbio mentor. Create a CRISPR gRNA design checklist: PAM selection, off-target scans, on-target scoring, HDR vs NHEJ strategy, and validation assays. Include a lab notebook template.

Act as a quality lead. Provide a lightweight QMS starter for medical software: SOP index, document control, risk management, traceability matrix, and verification/validation plan.

You are a regulatory strategist. Outline validation for software as a medical device/AI decision support: intended use, clinical association, analytical validation, clinical validation, and real-world monitoring.