Skip to content

Advertisement

You're viewing the new version of our site. Please leave us feedback.

Learn more

BMC Bioinformatics

Open Access

N-terminal sequence-based prediction of subcellular location

  • Evangelia I Petsalakis1Email author,
  • Pantelis G Bagos1,
  • Zoi I Litou1 and
  • Stavros J Hamodrakas1
BMC Bioinformatics20056(Suppl 3):S11

https://doi.org/10.1186/1471-2105-6-S3-S11

Published: 21 September 2005

Different compartments in a cell perform diverse tasks, thus knowledge of the localization of a protein would be highly indicative of its function. Many proteins have an N-terminal sequence of approximately 20–50 residues, which is responsible for their targeting to the appropriate location and is cleaved off, after the protein has been inserted into the organelle. Such information can be used in computational methods predicting protein localization. There are several methods for prediction of protein subcellular location available on the web, with TargetP (Emanuelsson et al 2000) being the most widely used. PredSL is a software tool, which aims to classify proteins to five subcellular locations: chloroplast, thylakoid, mitochondrion, secretory pathway and other. It combines neural networks, Markov chains, HMMs and scoring matrices in order to identify a targeting sequence at the N-terminal of a protein sequence, and determine its type (chloroplast-cTP, mitochondrial-mTP, secreted-SP, thylakoidal-lTP) and the precise location of the cleavage site. PredSL was tested on a set of 732 plant protein sequences and 637 non-plant sequences and the overall accuracy was 90.4% for the plant set and 93.3% for the non-plant set. Compared to the results obtained by TargetP when tested on the same datasets, PredSL's performance was better by 1.6% for the plant sequences and slightly worse (0.2%) for the non-plant sequences. For the prediction of thylakoid proteins PredSL was tested by cross-validation and achieved 87.3% accuracy compared to 87% by LumenP. PredSL is the only method for protein subcellular localization prediction that is available by the authors as a free, stand-alone tool, or through the URL: http://bioinformatics.biol.uoa.gr/PredSL.

Authors’ Affiliations

(1)
Department of Cell Biology and Biophysics, University of Athens Panepistimiopolis

Copyright

© Author(s); licensee BioMed Central Ltd. 2005

Advertisement