Abstract
During the structural determination of transmembrane proteins, one crucial piece of information is lost: the orientation of the protein within the lipid bilayer. The TmDet algorithm was developed in the early 2000s to determine the relative position of membrane proteins with respect to the lipid bilayer using the atomic coordinates of the protein. In the past 20 years, new types of transmembrane protein structures have emerged, such as the bacterial multidrug efflux pump, which is embedded in both the outer and inner membranes, or the piezo-type mechanosensitive ion channel, which is in a highly curved membrane. The breakthrough of artificial intelligence in protein structure prediction in recent years has also brought a new challenge to the field of transmembrane proteins; How can we determine the accuracy of a model structure if the actual three-dimensional structure of the protein is unknown? How can we determine the membrane orientation of a modeled transmembrane protein structure if the model contains errors? The new TmDet web server, available at https://tmdet.unitmp.org, provides answers to all these challenges, allowing researchers to determine how a transmembrane protein is in the membrane, to obtain information about the type of structural elements, and to determine whether the modeled structure can be embedded into the membrane.