Prompt Examples Overview

These examples are meant to teach the intended operating model of proteomics-mcp.

The main pattern is:

1. inspect or summarize an upstream ID/PSM table
2. select one peptide, PSM, or constrained subset
3. score, annotate, or render only the selected target
4. aggregate scored outputs only when needed

Each example includes:

• User prompt: what a user could ask an LLM or MCP-capable assistant
• What the MCP should do: the intended tool sequence
• Expected output shape: the kind of structured response or artifact the user should expect

Files

• 01_table_inspection.md
• 02_spectrum_level.md
• 03_batch_prediction_and_scoring.md
• 04_dataset_aggregation.md
• 05_sdrf_and_metadata.md
• 06_conversion_and_export.md
• 07_rendering_and_dashboards.md
• 08_troubleshooting.md

Guidance

Good prompts for this MCP usually:

• name the table or spectrum source
• indicate the adapter when obvious (instanovo, fragpipe, maxquant, proteome_discoverer)
• state whether the goal is inspection, scoring, rendering, or aggregation
• constrain the target set (top 10, one peptide, one run, one scan, one PRIDE accession)

Less useful prompts are vague requests like:

• "analyze this proteomics file"
• "score everything"
• "make all plots"

Those are underspecified and usually not the right operating pattern for this server.