[Membrane probability profile construction based on amino acids sequences multiple alignment]

Mol Biol (Mosk). 2006 May-Jun;40(3):541-5.
[Article in Russian]

Abstract

Prediction of membrane segments in sequences of membrane proteins is well known and important problem. Accuracy of the solution of this problem by methods that don't use homology search in additional data bank can be improved. There is a lack of testing data in this area because of small amount of real structures of membrane proteins. In this work, we create a testing set of structural alignments of membrane proteins, in which positioning of the membrane segments reflects agreement of known 3D-structures of proteins in the alignment. We propose a method for predicting position of membrane segments in multiple alignment based on forward-backward algorithm from HMM theory. This method not only allows to predict positions of membrane segments but also forms probability membrane profile, which can be used in multiple alignment methods that take into account secondary structure information about sequences. Method is implemented in computer program available on the World-Wide Web site http://bioinf.fbb.msu.ru/fwdbck/. Proposed method provides results better than MEMSAT method, which is nearly only tool for prediction of membrane segments in multiple alignments without additional homology search.

Publication types

  • Comparative Study
  • English Abstract

MeSH terms

  • Animals
  • Databases, Protein
  • Humans
  • Membrane Proteins / genetics*
  • Predictive Value of Tests
  • Protein Structure, Secondary
  • Protein Structure, Tertiary
  • Sequence Alignment*
  • Sequence Analysis, Protein*
  • Software*

Substances

  • Membrane Proteins