Skip to content

SKILL.md

Latest commit

 

History

History
180 lines (129 loc) · 5.46 KB

SKILL.md

File metadata and controls

180 lines (129 loc) · 5.46 KB

name: 'microbiome-cancer-agent' description: 'AI-powered analysis of microbiome-cancer interactions including tumor microbiome profiling, immunotherapy response prediction, and microbiome-targeted therapeutic opportunities.' measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes. allowed-tools:

  • read_file
  • run_shell_command

Microbiome-Cancer Interaction Agent

The Microbiome-Cancer Interaction Agent analyzes relationships between the microbiome and cancer, including tumor-associated bacteria, gut microbiome effects on immunotherapy, and microbiome-targeted therapeutic strategies.

When to Use This Skill

  • When analyzing tumor microbiome composition from sequencing data.
  • To predict immunotherapy response based on gut microbiome profiles.
  • For identifying microbiome-based biomarkers in cancer.
  • When assessing antibiotic impact on cancer treatment efficacy.
  • To design microbiome-modulating therapeutic interventions.

Core Capabilities

  1. Tumor Microbiome Analysis: Profile intratumoral bacteria from tumor sequencing data.

  2. Gut-Cancer Axis: Analyze fecal microbiome associations with cancer outcomes.

  3. ICI Response Prediction: Predict checkpoint inhibitor response from microbiome.

  4. Metabolite Profiling: Link microbial metabolites to cancer phenotypes.

  5. Antibiotic Impact: Model antibiotic effects on treatment efficacy.

  6. FMT/Probiotic Design: Support microbiome-modulating interventions.

Microbiome-Cancer Associations

Cancer Type Key Bacteria Association
Colorectal Fusobacterium nucleatum Promotion, poor prognosis
Colorectal Bacteroides fragilis (ETBF) Carcinogenesis
Gastric Helicobacter pylori Established carcinogen
Pancreatic Gammaproteobacteria Drug metabolism
Breast Fusobacterium Metastasis
Oral Porphyromonas gingivalis Oral SCC

Workflow

  1. Input: 16S/shotgun metagenomics, tumor sequencing, clinical data.

  2. Taxonomy Profiling: Identify bacterial composition at genus/species level.

  3. Diversity Analysis: Calculate alpha and beta diversity metrics.

  4. Association Testing: Correlate microbiome with outcomes.

  5. Functional Prediction: Infer metabolic potential (PICRUSt2, HUMAnN).

  6. Prediction Modeling: Build response prediction models.

  7. Output: Microbiome profile, associations, predictions, interventions.

Example Usage

User: "Analyze gut microbiome from melanoma patients and predict anti-PD-1 response."

Agent Action:

python3 Skills/Microbiome/Microbiome_Cancer_Agent/microbiome_cancer.py \
    --metagenomics fecal_shotgun.fastq.gz \
    --tumor_data melanoma_rnaseq.tsv \
    --clinical treatment_outcomes.csv \
    --analysis ici_response \
    --reference metaphlan_db \
    --output microbiome_report/

ICI Response and Microbiome

Favorable Microbiome:

  • Akkermansia muciniphila
  • Faecalibacterium prausnitzii
  • Bifidobacterium spp.
  • Ruminococcaceae family
  • High diversity

Unfavorable Microbiome:

  • Bacteroidales (in some studies)
  • Low diversity
  • Post-antibiotic dysbiosis

Microbial Metabolites in Cancer

Metabolite Source Effect
Butyrate Clostridia Anti-inflammatory, anti-tumor
Inosine Akkermansia Enhanced ICI response
TMAO Various Pro-tumorigenic
Secondary bile acids Various Variable, context-dependent
LPS Gram-negative Inflammation, mixed effects

AI/ML Components

Response Prediction:

  • Random forest on microbiome features
  • Neural networks for metagenomic profiles
  • Integration with host factors

Microbiome-Metabolite Linking:

  • Genome-scale metabolic models
  • Correlation networks
  • Causal inference methods

Intervention Design:

  • FMT donor selection
  • Probiotic consortium optimization
  • Antibiotic avoidance recommendations

Tumor Microbiome Analysis

Challenges:

  • Low bacterial biomass in tumors
  • Contamination from reagents/environment
  • Batch effects
  • Need for stringent controls

Best Practices:

  • Negative controls (extraction, PCR)
  • Decontamination algorithms (decontam, SCRuB)
  • Multiple validation methods
  • Orthogonal confirmation (FISH, culture)

Clinical Implications

  1. Biomarker Development: Microbiome-based response prediction
  2. Intervention Timing: Avoid antibiotics pre-ICI
  3. FMT Trials: Responder microbiome transfer
  4. Probiotics: Rationally designed consortia
  5. Prebiotics: Fiber to support beneficial bacteria

Prerequisites

  • Python 3.10+
  • QIIME2, Metaphlan, HUMAnN
  • R (phyloseq, vegan)
  • ML frameworks

Related Skills

  • Metagenomics - For general microbiome analysis
  • Immune_Checkpoint_Combination_Agent - For ICI optimization
  • Metabolomics - For metabolite analysis

Research Frontiers

  1. Intratumoral bacteria: Direct tumor effects
  2. Phage therapy: Targeting pathobionts
  3. Engineered probiotics: Drug-producing bacteria
  4. Diet interventions: Modulating microbiome for therapy

Author

AI Group - Biomedical AI Platform