G-protein-coupled receptors (GPCRs) are the largest family of membrane proteins in cellular signaling and targets of over one-third of approved drugs. Most drugs bind the conserved orthosteric site, often leading to limited subtype selectivity and dose-dependent side effects. Allosteric modulation provides an alternative by targeting distinct, less conserved sites, enabling more precise control of receptor activity. However, identifying these sites is difficult due to their transient and dynamic nature.This project aims to develop and validate a computational framework for systematic identification of GPCR allosteric sites. The approach combines molecular dynamics (MD) simulations of membrane-embedded receptors with information-theory analysis. Correlated residue motions will be quantified using mutual information and the Maximum Information Spanning Tree (MIST) algorithm to reveal communication networks linking distant regions with the orthosteric and G-protein binding sites. Dynamically coupled residues will be mapped onto structures to identify potential allosteric sites, which will then be assessed for druggability using fragment-based hotspot mapping.The pipeline will be optimized and benchmarked on class A GPCRs with known allosteric modulators, then extended to classes B, C, and F to evaluate general applicability. Finally, it will be applied to the adenosine receptor family to predict novel allosteric pockets and ligands.The outcome will be a robust, transferable, and openly available computational protocol for GPCR allosteric site detection, supporting experimental efforts and accelerating the discovery of more selective and safer drugs.