transFold

The transFold web server is a remarkable software tool that has been designed to predict the super secondary structure of transmembrane beta-barrel (TMB) proteins. These proteins are found in several cellular locations and are considered an essential protein class. However, the experimental difficulty in crystallizing transmembrane proteins has made it hard to determine their structures by X-ray diffraction. In light of this issue, transFold has emerged as a powerful solution.

What is particularly impressive about transFold is that it does not use machine learning methods such as hidden Markov models or neural networks, unlike previous approaches. Instead, the software tool employs multi-tape S-attribute grammars to describe all potential conformations. Afterward, it applies dynamic programming to determine the global minimum energy super secondary structure. This makes transFold unique and highly effective in predicting the structure of transmembrane proteins.

The transFold web server can predict not only the secondary structure and TMB topology but also the side-chain orientation of transmembrane beta-strand residues, inter-strand residue contacts, and TM beta-strand inclination with respect to the membrane. This is a significant development that has helped to improve the accuracy of beta-barrel structure prediction significantly.