Advancing cancer driver gene detection via Schur complement graph augmentation and independent subspace feature extraction

Xinqian Ma, Zhen Li, Zhenya Du, Yan Xu, Yifan Chen, Linlin Zhuo, Xiangzheng Fu, Ruijun Liu

Abstract: Highlights•We introduces a CDG prediction model that utilizes Schur complement graph argumentation and independent subspace learning.•We propose a strategy based on the Schur complement to generate enhanced views that capture critical topological structures.•We develop a feature extraction technique utilizing independent subspaces to improve the model's expressive capacity.•We incorporated a multi-layer attention mechanism into the GNN model to highlight the features most pertinent to the task.