agentleFS
aws-samplesaws-samples/amazon-bedrock-agents-healthcare-lifesciences/skills/biomarker-database-analysis/SKILL.md

biomarker-database-analysis

Skill274 starsChanged 6 months agoMIT-0View on GitHub

Use when a researcher needs to query biomedical databases for biomarker discovery, build target profiles from UniProt/Open Targets/STRING, rank biomarker candidates by evidence strength, or generate SQL queries against clinical genomic databases.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Always retrieve the database schema first to understand available tables and columns.

    line 25
  2. Distinctive:

    Write queries as single lines (no newlines)

    line 48
  3. Distinctive:

    Never modify column names from the schema

    line 49
  4. Distinctive:

    Use aggregation (COUNT, AVG, GROUP BY) to reduce output size

    line 50
  5. Very specific:

    Always validate with refine_sql before execution

    line 51
  6. Distinctive:

    Always include quality filters and use parameterized queries when available

    line 112
  7. Distinctive:

    When results exceed 100 rows, summarize with aggregation before presenting to user

    line 114

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Biomarker Database Analysis

When to use this skill

  • Researcher asks to find biomarkers associated with a disease or cancer type
  • Query clinical genomic databases for survival, gene expression, or mutation data
  • Build protein/target profiles from biomedical databases
  • Rank biomarker candidates by statistical evidence (p-value, effect size)
  • Generate or optimize SQL for biomarker data retrieval

MCP Servers Used

  • biomni-research — for external biomedical database queries (UniProt, Open Targets, STRING, ClinVar)
  • Clinical genomic databases may use separate tools (Redshift/Athena) depending on deployment

Workflow: Query Clinical Genomic Database

Step 1: Understand the schema before querying

Always retrieve the database schema first to understand available tables and columns.

Tool: get_schema
Purpose: Retrieve table names, column names, data types, and descriptions

Key columns in a typical clinical genomic table:

  • case_id -- patient identifier
  • survival_status -- alive/dead (boolean or 0/1)
  • survival_duration -- time in days or years
  • Gene expression columns (e.g., gdf15, lrig1, cdh2, postn, vcan)
  • Clinical metadata: age_at_histological_diagnosis, smoking_status, chemotherapy, histology

Step 2: Formulate and refine the SQL query

Decision tree for query type:

  • Patient demographics -> Simple SELECT with WHERE/GROUP BY
  • Biomarker expression -> SELECT gene columns with clinical filters
  • Survival correlation -> SELECT survival_status, survival_duration, biomarker columns
  • Cohort comparison -> GROUP BY with aggregation (COUNT, AVG)

Rules:

  1. Write queries as single lines (no newlines)
  2. Never modify column names from the schema
  3. Use aggregation (COUNT, AVG, GROUP BY) to reduce output size
  4. Always validate with refine_sql before execution
Tool: refine_sql
Input: sql (the query), question (rationale for this step -- not the user's original question)
Purpose: Optimize for efficiency, add aggregation, fix column references

Step 3: Execute and interpret results

Tool: query_redshift (or query_database)
Input: The refined SQL query
Output: Row-level results from the clinical database

Step 4: Build target profiles from external databases

For deeper biomarker validation, use the biomni-research MCP server with natural language queries:

DatabaseQuery approach
UniProt"CDK4 protein function, domains, post-translational modifications"
Open Targets"CDK4 disease associations and genetic evidence scores"
STRING"CDK4 protein-protein interaction network"
ClinVar"CDK4 pathogenic variants and clinical significance"

Step 5: Rank candidates by evidence strength

Scoring framework for biomarker prioritization:

Evidence typeWeightSource
Statistical significance (p < 0.05)HighCox regression from clinical data
Known pathogenic associationHighClinVar, Open Targets
Protein interaction in disease networkMediumSTRING (confidence > 0.7)
Literature support (3+ publications)MediumPubMed
Gene expression differentialMediumClinical database
Functional annotation matchLowUniProt

Query Patterns

Find top biomarkers for survival:

SELECT survival_status, survival_duration, gdf15, lrig1, cdh2, postn, vcan FROM clinical_genomic WHERE chemotherapy = 'Yes'

Cohort demographics:

SELECT smoking_status, COUNT(DISTINCT case_id) AS num_patients FROM clinical_genomic WHERE age_at_histological_diagnosis > 50 GROUP BY smoking_status

Disease-specific expression:

SELECT survival_status, COUNT(*) AS count FROM clinical_genomic WHERE histology = 'Adenocarcinoma' GROUP BY survival_status

Key Conventions

  • Map survival_status: False/Alive = 0, True/Dead = 1
  • Expression values are continuous (higher = more expressed in tumor)
  • Always include quality filters and use parameterized queries when available
  • Store query results in shared memory for downstream agents (statistician, pathway analyst)
  • When results exceed 100 rows, summarize with aggregation before presenting to user

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.