Plasmatic membranes contain a high amount of membrane proteins. They perform vital functions of life, so any disruptions in their structure result in pathologies and diseases. Studies of these proteins with experimental methods are very complicated and expensive, as they require the membrane environment. Despite considerable progress achieved so far in methods of structure determination and property analysis, many computationalmethods are developing to predict the structural and dynamical parameters of proteins in membranes. Among the algorithms of modeling are the homology analysis, de novo structure prediction, molecular dynamics simulations and other. With growing computational capabilities, sophisticated techniques are developed taking into account more environmental factors. Combined approaches with different levels of approximation of intermolecular interactions are widely used. The major interest in studies of membrane proteins is focused on their transmembrane domains that are fundamental structural elements and are constituted by α-helices or helical bundles incorporated into lipid bilayer in most cases. Therefore, the fundamental problem of interaction of a pair of helices in membrane arises: the exact mechanism of this process is still not so clear. In place of the prevailing concept of dimerization motifs that states the importance of protein-protein contacts, a new model of the membrane as an adaptable lipid matrix is proposed. It states that biological membrane can adjust its properties around proteins and also modulates their activity. This mechanism of the mutual influence of two components is challenging modern computationalmethods of membrane modeling because these systems are quite large and includemany components to be treated accurately. Nowadays, investigations of the complex multi-component model systems become possible with modern methods of computational experiment.