EARN: an ensemble machine learning algorithm to predict driver genes in metastatic breast cancer

Area of Science:

• Genomics
• Computational Biology
• Oncology

Background:

• Limited understanding of drivers influencing cancer aggression in complex diseases like breast cancer.
• Need for better prognostic and diagnostic markers for primary and metastatic breast cancer.

Purpose of the Study:

• To identify plausible driver genes for metastatic breast cancer (MBCA) using somatic mutation data.
• To develop an ensemble classifier (EARN) for evaluating driver genes.
• To propose a novel gene set panel for MBCA prognosis and diagnosis.

Main Methods:

• Analysis of 450 metastatic breast tumor samples from cBio Cancer Genomics Portal.
• Feature extraction using four software tools.
• Development and application of an ensemble classifier (EARN) combining Artificial Neural Network, Random Forest, and non-linear Support Vector Machine.
• Gene set enrichment analysis and pathway enrichment analysis (PEA) using ReactomeFIVIz.

Main Results:

• Identification of predicted driver and passenger genes.
• Proposed a novel gene set panel for MBCA including HDAC3, ABAT, GRIN1, PLCB1, KPNA2, NCOR1, TBL1XR1, SIRT4, KRAS, CACNA1E, PRKCG, GPS2, SIN3A, ACTB, KDM6B, and PRMT1.
• Achieved high ROC-AUC values (99.24% for MBCA, 99.79% for BRCA) using EARN, outperforming individual classifiers.

Conclusions:

• The integrative approach assists precision oncologists in designing targeted gene panels.
• Eliminates the need for whole-genome/exome sequencing for certain applications.
• Provides a foundation for improved diagnostic and therapeutic strategies in breast cancer.