ExGenet, Integrating Design of Experiments and Response Surface Methodology for Cancer Gene Detection in Gene...

Network analysis techniques often require tuning hyperparameters for optimal performance. For instance, the independent cascade model necessitates determining the probability of diffusion. Despite its importance, a consensus on effective parameter adjustment remains elusive.

In this study, we propose a novel approach utilizing experimental design methodologies, specifically 2-Factorial Analysis for Screening, and Response Surface Methodology (RSM) for parameter adjustment. We apply this methodology to the task of detecting cancer driver genes in colorectal cancer.

Through experimental validation of colorectal cancer data, we demonstrate the effectiveness of our proposed methodology. Compared with existing methods, our approach offers several advantages, including reduced computational overhead, systematic parameter selection grounded in statistical theory, and improved performance in detecting cancer driver genes.

This study presents a significant advancement in the field of network analysis by providing a practical and systematic approach to hyperparameter tuning. By optimizing parameter settings, our methodology offers promising implications for critical biomedical applications such as cancer driver gene detection.