Evidence-Aware Protein Complex Detection: Methods, Benchmarks, and Reproducibility Challenges

This arXiv feed collects method papers across graph methods, machine learning for science forecasting, social-media auditing, and computational biology. Notable items with security/abuse relevance include: SAHG, a sector-anisotropic hyperbolic-geometry graph model aimed at improving social-bot detection via relational signals; TAGR, a topology-aware Gaussian graph-repair method to make GNNs robust to noisy/missing edges; and a fast curvature approximation (CCOM) for Ollivier–Ricci curvature useful in community detection and network analysis. Other papers cover explainable forecasting of emerg[