PathMethy

PathMethy traces the origin of tumors in cancer of unknown primary (CUP) samples based on DNA methylation using a Transformer model integrated with functional categories and crosstalk of pathways.

Key Features:

• Cancer Origin Tracing: Accurately traces the origins of both primary and metastatic tumors in CUP samples.
• Transformer Architecture: Employs a Transformer model integrated with functional categories and crosstalk of pathways for prediction.
• Pathway Interpretability: Provides biological insights by highlighting key pathways, functional categories, and their interactions.
• Molecular Subtype Prediction: Predicts molecular subtypes within nine primary tumor types.
• Global Biological Process Understanding: Uses functional categories of pathways to gain a global understanding of biological processes.

Scientific Applications:

• Cancer of Unknown Primary Diagnosis: Diagnoses the tissue of origin for CUP cases, which represent 3%-5% of cancer cases that remain unclassified despite advanced diagnostics.
• Metastatic Tumor Origin Identification: Identifies the primary origin of metastatic tumors based on DNA methylation patterns.
• Cross-Dataset Benchmarking: Outperformed seven competing methods in F1-score across nine cancer datasets.
• Clinical Validation: Demonstrated high agreement with previously diagnosed sites in CUP cases.

Methodology:

PathMethy uses a Transformer model integrated with functional categories and crosstalk of pathways, trained on DNA methylation data to trace tumor origins and predict molecular subtypes.