Atomic-level information is essential to describe the structure and dynamics of biomolecular assemblies. The work presented in this thesis aims to explore and enhance computational techniques explaining the formation of complexes, quantifying binding free energies or describing the dynamics of multi-components systems. I first developed a protocol to compute the binding free energy of a ligand buried in a membrane protein. It relies on alchemical transformations carried out in a rigorous statistical mechanical framework. The protocol is distributed within the BFEE2 plugin, a tool designed to assist the end user in preparing all the necessary input files and performing the posttreatment of the simulations towards the final estimate of the binding affinity. Molecular Dynamics (MD) and alchemical simulations have been employed to provide insights into the formation of specific protein complexes in terms of structure and dynamics. The set of Dpr and DIP proteins, which play a key role in the neuromorphogenesis in the nervous system of Drosophila melanogaster, offer a rich paradigm to learn about proteinprotein recognition. Many members of the DIP subfamily cross-react with several members of the Dpr family and vice-versa. While there exists a total of 231 possible Dpr-DIP heterodimer, only 57 “cognate” pairs have been detected by surface plasmon resonance (SPR) experiments, suggesting that the remaining 174 pairs have low or unreliable binding affinity. Here I assessed the performance of computational approaches to in quantifying the binding affinities between Dpr and DIP proteins and I identified by means of a series of point mutations, the interfacial residues governing the specificity of the recognition process. Building on alchemical transformations, I developed a hybrid nonequilibrium molecular dynamics - Monte Carlo (neMD/MC) simulation method to aimed at enhancing the sampling of inhomogeneous membranes, circumventing the slow lateral diffusion of the various constituents. Randomly chosen lipid molecules are swapped to generate configurations that are subsequently accepted or rejected according to a Metropolis criterion based on the alchemical iii work associated to the attempted swap calculated via a short trajectory. The performance of the hybrid neMD/MC algorithm and its ability to sample the distribution of lipids near a transmembrane helix carrying a net charge are illustrated for a binary mixture of charged and zwitterionic lipids. To enforce equilibrium between a simulated system and an infinite surrounding bath, a modified version of the neMD/MC algorithm was developed, in which a randomly chosen lipid molecule in the simulated system is swapped with a lipid picked in a separate system standing as a thermodynamic “reservoir” with the desired mole fraction for all lipid components. Membrane proteins function has been shown to depend on the lipid organization within the membrane either through averaged bulk effect or specific binding. A well-known class of protein exhibiting such a dependance is the family of pentameric ligand-gated ion channels (pLGICs). Upon the binding of a neurostransmitter, the conformation of these proteins changes establing a ionic current at the synapse junctions, transforming therby a chemical into an electric signal. Here, we generated several MD trajectories of various agonist-bound structures of nicotonic acethlycholine receptors solved by cryoEM, providing a molecular basis shedding light on the desensitization process. The conductivity and the stability of the pore of the pLGICs in a desensitized state are measured. The functions of these proteins have also been shown to depend on lipid composition. Finally, we employed alchemical tranformations to quantify the relative binding affinities of anionic and zwitterionic lipids at putative pLGIC binding sites, enlightening how lipids modulate the fonction of these proteins.