Skip to content

SKILL.md

Latest commit

 

History

History
182 lines (129 loc) · 5.23 KB

SKILL.md

File metadata and controls

182 lines (129 loc) · 5.23 KB

name: 'cellfree-rna-agent' description: 'AI-powered cell-free RNA analysis from liquid biopsy for cancer detection, tissue-of-origin identification, and non-invasive transcriptomic profiling.' measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes. allowed-tools:

  • read_file
  • run_shell_command

Cell-Free RNA Analysis Agent

The Cell-Free RNA Analysis Agent provides comprehensive analysis of circulating cell-free RNA (cfRNA) from plasma and other biofluids for cancer detection, tissue-of-origin identification, and non-invasive transcriptomic profiling.

When to Use This Skill

  • When analyzing plasma cfRNA for cancer detection and monitoring.
  • To identify tissue-of-origin from circulating transcripts.
  • For non-invasive transcriptomic profiling of tumors.
  • When integrating cfRNA with cfDNA for comprehensive liquid biopsy.
  • To discover RNA-based biomarkers from accessible biofluids.

Core Capabilities

  1. cfRNA Profiling: Quantify mRNA, lncRNA, and small RNA from plasma.

  2. Tissue Deconvolution: Identify tissue sources contributing to cfRNA pool.

  3. Cancer Detection: ML models detecting cancer from cfRNA profiles.

  4. Tumor Transcriptomics: Infer tumor gene expression non-invasively.

  5. Integration with cfDNA: Combine RNA and DNA liquid biopsy analytes.

  6. Biomarker Discovery: Identify diagnostic and prognostic RNA markers.

cfRNA Biology

Sources:

  • Cell death (apoptosis, necrosis)
  • Active secretion (EVs, RNA-binding proteins)
  • Cell surface-associated RNA

Protection Mechanisms:

  • Extracellular vesicles
  • Protein complexes (AGO2, NPM1)
  • Lipoproteins

Half-life: Minutes to hours (shorter than cfDNA)

Workflow

  1. Input: Plasma cfRNA sequencing data (total RNA, small RNA, or targeted).

  2. Quality Control: Assess library complexity, mapping rates, contamination.

  3. Quantification: Normalize and quantify transcripts.

  4. Deconvolution: Estimate tissue contributions.

  5. Classification: Apply cancer detection models.

  6. Integration: Combine with cfDNA if available.

  7. Output: Tissue composition, cancer score, biomarker profiles.

Example Usage

User: "Analyze plasma cfRNA to detect cancer and identify tissue of origin."

Agent Action:

python3 Skills/Genomics/CellFree_RNA_Agent/cfrna_analyzer.py \
    --input plasma_cfrna.fastq.gz \
    --protocol total_rna \
    --reference gencode_v44 \
    --deconvolution true \
    --cancer_detection true \
    --output cfrna_results/

Tissue Deconvolution

Reference Transcriptomes:

  • GTEx tissue expression atlas
  • Single-cell reference atlases
  • Tissue-specific marker genes

Methods:

  • Non-negative least squares
  • Support vector regression
  • Deep learning deconvolution

Clinical Applications:

  • Organ injury detection (liver, heart, brain)
  • Tumor burden estimation
  • Post-transplant monitoring

Cancer Detection Applications

Cancer Type Key Markers Performance
Lung XIST, MALAT1, specific mRNAs AUC 0.80-0.90
Breast HER2, ER/PR transcripts Monitoring
Colorectal KRAS, panel genes Early detection
Prostate PCA3, TMPRSS2-ERG Established
Liver AFP, specific ncRNAs HCC surveillance

Technical Considerations

Pre-analytical Factors:

  • Sample collection (EDTA, cell stabilization)
  • Processing time (<4 hours recommended)
  • Storage temperature (-80°C)
  • Hemolysis avoidance (critical)

Library Preparation:

  • Total RNA (captures mRNA, lncRNA)
  • Small RNA (miRNA, piRNA)
  • Targeted panels (specific genes)
  • UMI-based for quantification

AI/ML Components

Cancer Classifier:

  • Gradient boosting on gene panels
  • Neural networks for full transcriptome
  • Multi-cancer detection models

Tissue Predictor:

  • Reference-based deconvolution
  • Supervised tissue classifiers
  • Anomaly detection for novel sources

Integration with Other Analytes

Analyte Strength Combination Benefit
cfDNA Mutations, methylation Genomic + transcriptomic
CTCs Single-cell analysis Cellular confirmation
Exosomes Protected RNA Source identification
Proteins Functional markers Multi-modal biomarkers

Prerequisites

  • Python 3.10+
  • STAR/Salmon for alignment
  • DESeq2/edgeR for quantification
  • Tissue deconvolution tools

Related Skills

  • Liquid_Biopsy_Analytics_Agent - For comprehensive liquid biopsy
  • Exosome_EV_Analysis_Agent - For EV-derived RNA
  • ctDNA_Analysis - For DNA-based markers

Emerging Technologies

  1. Targeted cfRNA: Gene panels for specific cancers
  2. Single-molecule: Direct RNA sequencing
  3. Spatial deconvolution: Mapping cfRNA to tissue regions
  4. Longitudinal monitoring: Treatment response tracking

Author

AI Group - Biomedical AI Platform