Get access

Prediction of the human membrane proteome

Authors

  • Linn Fagerberg,

    1. School of Biotechnology, AlbaNova University Center, Royal Institute of Technology (KTH), Stockholm, Sweden
    Search for more papers by this author
  • Kalle Jonasson,

    1. School of Biotechnology, AlbaNova University Center, Royal Institute of Technology (KTH), Stockholm, Sweden
    Search for more papers by this author
  • Gunnar von Heijne,

    1. Department of Biochemistry and Biophysics, The Arrhenius Laboratories for Natural Sciences, Stockholm University, Stockholm, Sweden
    Search for more papers by this author
  • Mathias Uhlén,

    1. School of Biotechnology, AlbaNova University Center, Royal Institute of Technology (KTH), Stockholm, Sweden
    Search for more papers by this author
  • Lisa Berglund

    Corresponding author
    1. School of Biotechnology, AlbaNova University Center, Royal Institute of Technology (KTH), Stockholm, Sweden
    • School of Biotechnology, AlbaNova University Center, Royal Institute of Technology (KTH), 106 91 Stockholm, Sweden Fax: +46-8-5537-8482
    Search for more papers by this author

Abstract

Membrane proteins are key molecules in the cell, and are important targets for pharmaceutical drugs. Few three-dimensional structures of membrane proteins have been obtained, which makes computational prediction of membrane proteins crucial for studies of these key molecules. Here, seven membrane protein topology prediction methods based on different underlying algorithms, such as hidden Markov models, neural networks and support vector machines, have been used for analysis of the protein sequences from the 21 416 annotated genes in the human genome. The number of genes coding for a protein with predicted α-helical transmembrane region(s) ranged from 5508 to 7651, depending on the method used. Based on a majority decision method, we estimate 5539 human genes to code for membrane proteins, corresponding to approximately 26% of the human protein-coding genes. The largest fraction of these proteins has only one predicted transmembrane region, but there are also many proteins with seven predicted transmembrane regions, including the G-protein coupled receptors. A visualization tool displaying the topologies suggested by the eight prediction methods, for all predicted membrane proteins, is available on the public Human Protein Atlas portal (www.proteinatlas.org).

Get access to the full text of this article

Ancillary