Sosui is a freely accessible online program calculated from a user entered from the amino acid sequence of a specific portion of the secondary structure of proteins, the α - helix. The main task of the program, however, is to predict based on certain properties of the found α -helices with a relatively high degree of certainty whether the predetermined amino acid sequence of a soluble protein or a transmembrane protein.


The algorithm of Sosui was developed in 1996 at the University of Tokyo, the name in Japanese means, inter alia, "(his ) hydrophobic" and alluding to the "victims " of the program.


First Sosui searches for α - helices, which can be taking into account the well-known " helix potentials " (see article α - helix) of the amino acids is relatively easy to predict. This is followed by the much more challenging task of distinguishing between α -helices in soluble proteins and α -helices in transmembrane proteins. Sosui bases its calculation on four properties of the amino acid sequence:

A decisive step forward from the sole consideration of the hydropathy is the introduction of so-called " amphiphilicity index", which is calculated by each AS with a certain amphiphilic rest, those resulting from the molecular structure value is attributed. In order for Sosui to be amphiphilic, the hydrophilic polar residue must not depend directly on the β -carbon atom; it must rather have at least one non-polar carbon atom interposed be ( it follows that lysine, arginine, histidine, glutamate, glutamine, tryptophan, and tyrosine for the " amphiphilicity index" are relevant). The distinction between transmembrane α - helices and the remaining widely used α -helices able Sosui by attaching the two ends of the transmembrane α -helices detects the characteristic " accumulation " amphiphilic AS ( which provides in vivo that these α - helices transmembrane positioning the ends of the lipid - water interface is energetically favored ). The charge of the AS is also included in the calculation as their length; this Sosui exploits the fact that biological lipid membranes have a certain thickness ( about 8 nm) and transmembrane domains must therefore be a similar length.

This approach allowed a differentiation in 99% of cases examined in a study proteins. However, this study was conducted by the Sosui - makers themselves; in practice Sosui not reach such a high hit rate. One independent study, in which the number of transmembrane α -helices of 122 known transmembrane proteins was predicted with α -helix of different programs, Sosui was right in nearly 60 % of cases. Thus, even if the number is often predicted inaccurately but possible to distinguish between transmembrane proteins and soluble proteins mostly ( as it is sufficient for this purpose to find out if a protein in general a transmembrane α -helix has ). Of course, no membrane proteins can thus be identified which are not anchored to a transmembrane α -helix, for example, Porins or covalently bound to a lipid molecule proteins.


Sosui shows on the results page at the top of some general information such as length and average hydrophobicity. If it is a transmembrane protein, the number of the transmembrane domains and the location is specified. Then follows a hydropathy profile with highlighting the hydrophobic moieties, including in transmembrane proteins, the helical wheel ( "helical wheel" ) diagrams with color marking of polar or charged side chains AS. Finally, a schematic overview of the entire aa sequence is commanded, where the "primary" and "secondary" α -helices are indicated separately (as primarily an α -helix is called by the authors, when it is highly hydrophobic, being secondary if it has hydrophilic portions and thus can not be reliably classified as a transmembrane domain solely by the degree of their hydrophobicity, but only in consideration of the remaining three properties).