Effective Resistance and GNN Reliability in Tissue-Specific Interactomes
What it is
This paper investigates whether effective resistance in tissue-specific protein interaction networks serves as a reliable signal for predicting the uncertainty of Graph Neural Network (GNN) predictions. It finds that while effective resistance is largely dominated by node degree, the residual variance beyond degree correlates with model loss, suggesting it captures specific structural reliability information.
The problem it solves
Protein function annotation models often lack a reliable way to distinguish between confident and uncertain predictions, leading to potential errors in downstream biological applications.
What you could use it for
- Identifying unreliable nodes in protein function prediction models
- Analyzing structural properties of tissue-specific interactomes
- Improving selective prediction in biological graph neural networks
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.