Q2. How does the program determine transmembrane helices?
A2. To idenify transmembrane helices, the SOSUI system (1) employs two kinds of physical-chemical indices (an index by Kyte and Doolittle (2), and a new amphiphilicity index proposed by us). Predicted transmembrane helices by SOSUI are classified into "primary" and "secondary" classes. We call this classification "a multi-mechanism hypothesis, for formation of transmembrane helices", in which several physical-chemical factors are important in secondary-structural formation of transmembrane helices.
A "primary" transmembrane helix can be found out in the highly hydrophobic, successive region of your query sequence by a hydropathy profile based on the Kyte-Doolittle hydropathy index. Such transmembrane helices may consist mainly of apolar residues, and therefore may be inserted by theirselves into the hydrophobic lipid-bilayer. However, there will be less hydrophobic transmembrane helices, which originate from some polar amino-acid residues, than "primary" segments. They must be detected for better prediction of transmembrane helices.
We defined an amphiphilicity index of polar amino acids, to predict such less hydrophobic, "secondary" helices. Polar residues are often functionally significant in the interior of membrane proteins, for example, in case of ion pairs of channel proteins. Most of polar amino acids have an amphiphilic part attached to a long hydrocarbon chain or a bulky aromatic ring. If a polar residue of a transmembrane helix locates near the mebrane surface, its amphiphile would be contacted with water molecules or polar head-groups of lipids for the thermodynamically stable state.
If a polar residue exists in the core of a transmembrane helix, it could be explained in terms of the attractive polar interaction between transmembrane helices (3), (4). Our amphiphilicity index was determined as follows.
1) We estimated phase-transfer energy of each amino acid, supposed by the accessible surface area of the hydrocarbon part.
2) An amphiphilicity of each amino acid was calculated from the transfer energy above, by weighing with its intensity of polarity.
3) Charged and non-charged polar residues were exemplified.
In the SOSUI system, the profile based on the amiphiphilicity index is employed to elongate the primary helices, and search for the secondary ones. A non-charged polar residue is favorable for the secondary segments in our current predition.
If a segment in your query is beyond the "primary" threshold or the "secondary" threshold, it is predicted as a transmembrane helix. The current version of SOSUI puts an emphasis on descrimination of transmembrane helices, but determination of an exact transmembrane region. In the near future, the SOSUI system will be also improved concerning these points.
[References]
(1) Hirokawa, T., Seah, B-C., Mitaku, S.: Bioinformatics, 14 (4), 378-379 (1998)
(2) Kyte, J., Doolittle, R. F.: J. Mol. Biol., 157, 105-132 (1982)
(3) Mitaku, S., Suzuki, K., Odashima, S., Ikuta, K., Suwa, M., Kukita, F., Ishikawa, M., Itoh, H.: Proteins, 22, 350-362 (1995)
(4) Suwa, M., Hirokawa, T., Mitaku, S.: Proteins, 22, 363-377 (1995)