bHiTAIC

bHiTAIC classifies the tissue of origin and tumor type in primary and metastasized cancers using genome-scale DNA methylation profiles to enable tracing of tumor origins for diagnostic and research applications.

Key Features:

• Hierarchical Model: Employs a hierarchical classification model that leverages genome-scale DNA methylation data and a multilayer perceptron architecture to distinguish cancer-type–specific methylation patterns.
• High Accuracy and Resolution: Reported 99% precision for primary cancer origin identification on the test dataset, 93% accuracy on an external validation set for primary cancers, and 96% accuracy for metastatic tissue-of-origin identification in the external dataset.
• Comprehensive Coverage: Validated across 27 cancer types originating from 23 distinct tissue sites using data from 7,735 tumors.

Scientific Applications:

• Clinical Diagnosis: Identification of tissue of origin for cancers with unknown primary sites to inform cancer-type–specific treatment decisions.
• Research on Carcinogenesis: Tracing tumor origins via DNA methylation alterations to investigate carcinogenesis processes and cell-fate differentiation.
• Metastasis Studies: Determination of tissue origins of metastatic tumors to support studies of metastatic pathways and potential therapeutic targets.

Methodology:

Uses a hierarchical classification model trained on genome-scale DNA methylation profiles with a multilayer perceptron architecture and evaluated on internal test and external validation datasets.