Comparing Statistical Metrics for Differential Network Analysis

C. Ruiz de Porras Rubio, A. Sánchez Pla, C. Andrés-Lacueva, D. Garrido-Martín, A. Miñarro Alonso

Network analysis is a powerful tool in biology, medicine, and social sciences for studying complex systems, with nodes representing entities and edges their interactions. In this work, networks are constructed from partial correlation matrices, where edges represent associations between pairs of variables conditional on all others. Changes in networks help reveal pathological mechanisms and system responses to external stimuli. Differential network analysis compares structures under different conditions, detecting shifts in connections beyond individual node expression. This study evaluates statistical metrics on simulated weighted matrices using permutation tests, showing that Frobenius distance, Invariant Network Structure, L1-norm and Correlation distance effectively detect subtle structural differences. Applied to real data, the approach identifies significant changes between experimental conditions.

Keywords: Partial correlation networks Differential network analysis Permutation tests Distance metrics Weighted matrices Power analysis

Scheduled

Poster session II

September 4, 2026 9:00 AM

Facultade de Ciencias Económicas e Empresariais

Other papers in the same session

Análisis cuantitativo del sector nuclear en la Unión Europea y Reino Unido

M. M. Alvarado, M. Mateo-Doll, J. Bautista-Valhondo

Inferencia Estadística para Determinar la Calidad de Test Diagnósticos Binarios

M. Á. Montero Alonso, A. Ruiz Vega, J. M. Melchor Rodríguez, J. D. Luna del Castillo

Longitudinal Statistical Analysis and Phenotypic Characterization of a Salt-Wasting Tubulopathy Cohort: Gitelman and Bartter Syndromes

L. Núñez González, B. Temes Álvarez, A. Urisarri Ruiz de Cortázar, M. Gil Calvo, A. M. Barcia de la Iglesia, P. Oliveros Martínez, N. Carrera Cachaza, M. Á. García González, M. García Murias

Modelización de biomasa de especies invasoras mediante el uso de datos LIDAR y espectrales de alta resolución obtenidos a partir de drones

N. Montiel, C. Pérez-Carral, P. J. Hidalgo, M. Ortega-Moreno, M. Ortega Moreno

Modelo para el problema de corte con remanentes reutilizables en el habilitado industrializado de acero

L. Padrón Blanco

Spatial Generalization under Hierarchical Dependence: Evidence from Sediment Modelling

C. Patino Alonso, J. L. Molina González, F. Espejo Almodóvar, S. Zazo del Dedo, M. C. Molina González

Sufficiency and Completeness for Markov Kernels: Some Non-Trivial Examples

P. Perez Fernandez, A. García Nogales

Regresión cuantil non paramétrica con covariables composicionales

A. Pérez González, M. Conde Amboage

Imputación de datos faltantes y predicción conjunta en apicultura de precisión mediante modelos autorregresivos bayesianos

M. C. Robustillo, L. Naranjo, M. I. Parra Arévalo, J. Hernández, C. J. Pérez Sánchez