A novel algorithm for network-based prediction of cancer recurrence.

Area of Science:

• Computational biology
• Cancer genomics
• Epigenetics

Background:

• Developing accurate prognostic models is crucial in omics-based cancer research.
• Identifying reliable biomarkers for cancer recurrence prediction remains a significant challenge.

Purpose of the Study:

• To propose a novel computational method for identifying dysregulated gene subnetworks as biomarkers.
• To predict cancer recurrence using DNA methylome data.

Main Methods:

• Applied a novel computational method to the DNA methylome of endometrial cancer patients.
• Identified Epigenetic Connectors (ECs) topologically connecting differentially methylated (DM) genes in a protein-protein interaction network.
• Combined DM genes and ECs using a random walk procedure for feature selection.

Main Results:

• Identified a significant subnetwork of ECs enriched in tumorgenesis and metastasis pathways.
• Constructed a support vector machine classifier that significantly improved cancer recurrence prediction accuracy.
• The network-based approach outperformed several alternative prediction methods.

Conclusions:

• The proposed network-based computational method effectively identifies gene subnetworks for cancer recurrence prediction.
• Epigenetic Connectors play a vital role in linking differentially methylated genes for improved prognostic modeling.
• This approach demonstrates significant potential for advancing omics-based cancer research and clinical applications.