CurvEndo: Geometry-dependent reorganization of membrane-contact patches in peripheral membrane proteins

CurvEndo is a structure-based computational framework developed to quantify how membrane geometry reorganizes protein–membrane contact patterns from static protein structures. The framework evaluates protein proximity to analytical membrane geometries including plane, sphere, cylinder, and membrane neck surfaces at curvature radii of 15, 30, and 60 nm. The deposited package contains the reproducible computational workflow associated with the CurvEndo study, including analysis-ready protein structures, Curvature Response Fingerprint (CRF v0.1) outputs, Geometry-selective Contact Persistence (GCP v0.1) results, orientation-based null models, validation analyses, local curvature complementarity assessment, numerical summary tables, and Python scripts. The benchmark comprises ten peripheral membrane proteins representing FYVE, PH, C1B, FERM, N-BAR, F-BAR, ENTH, I-BAR, and BAR architectures. CurvEndo does not use molecular dynamics, docking, energy minimization, or binding free-energy calculations. The descriptors are intended to quantify geometry-dependent organization of membrane-proximal contact patches rather than predict binding affinity or experimentally validated curvature sensing. The dataset supports the associated manuscript: “Geometry-Dependent Reorganization of Membrane-Contact Patches in Peripheral Membrane Proteins” Randomized analyses were performed using fixed pseudorandom seeds to facilitate reproducibility.