🤖 AI Summary
This study addresses the lack of theoretical grounding for assessing the reliability of graph neural networks (GNNs) in protein function prediction, specifically investigating whether tissue-specific network structures can provide node-level prediction confidence. By leveraging effective resistance to quantify GNN reliability within tissue-specific interactomes, this work reveals its degeneration into inverse node degree. Through null model testing and a selective prediction framework, it analyzes the explanatory power of residual signals—obtained after removing degree effects—on prediction errors. The findings demonstrate that these residual signals independently explain node-wise loss, with their explanatory variance increasing monotonically with network depth and reaching 5.6 times that of permutation baselines. This research establishes the first topology-based node-level confidence metric, uncovering both the degree-degeneration limitation of effective resistance and the robustness of residual signals.
📝 Abstract
Protein function annotation needs to know which predictions to distrust, not only what a model predicts. We ask whether tissue-specific interaction structure carries that information. Our candidate signal is effective resistance, used previously to relieve over-squashing by rewiring. Across 24 tissue-specific interactomes it is dominated by inverse degree, and the degeneration deepens as the co-expression filtered network grows, with a Spearman correlation of -0.955. The residual departure from that limit exceeds degree-preserving null graphs in all 24 networks. Controlling for predictive entropy, degree, annotation cardinality, local structure and feature-only difficulty, the residual explains additional per-node loss in 19 of 24 held-out networks once a permutation floor is subtracted, at every depth, and the effect strengthens monotonically with depth. The increment reaches 0.37% of the variance the controls leave unexplained, 5.6 times a permutation floor, against 1.5 times when the model is retrained in a degree-preserving null world. Selective prediction improves negligibly. The signal is reproducible; degree degeneration bounds it.