Standard Article

Image Analysis Tools in Proteomics

  1. Andrew W Dowsey1,
  2. Jane A English2,
  3. Frederique Lisacek3,
  4. Jeffrey S Morris4,
  5. Guang-Zhong Yang1,
  6. Michael J Dunn2

Published Online: 17 JAN 2011

DOI: 10.1002/9780470015902.a0006216.pub2

eLS

eLS

How to Cite

Dowsey, A. W., English, J. A., Lisacek, F., Morris, J. S., Yang, G.-Z. and Dunn, M. J. 2011. Image Analysis Tools in Proteomics. eLS. .

Author Information

  1. 1

    Imperial College London, South Kensington, London, UK

  2. 2

    University College Dublin, Belfield, Ireland

  3. 3

    Swiss Institute of Bioinformatics, Geneva, Switzerland

  4. 4

    The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Publication History

  1. Published Online: 17 JAN 2011

Abstract

Image analysis tools have matured into a number of established commercial packages and freely available programs that underpin research in expression proteomics. With respect to the available tools, we describe the challenges and methods for image analysis in two-dimensional gel electrophoresis and the emerging high-throughput ‘shotgun’ proteomics platform of liquid chromatography coupled to mass spectrometry. Two workflows can be identified: the established pipeline of spot detection followed by spot matching, and the new alternative pipeline of image alignment followed by consensus spot detection. In practice, image analysis is often viewed as a major bottleneck in proteomics, and we will further touch on emerging research that aims to statistically model and integrate data earlier in the pipeline to minimise manual interaction and therefore the subjective bias and restrictions that it brings.

Key Concepts:

  • Precise quantification and differential analysis of protein expression is reliant on semiautomated image analysis tools.

  • Liquid chromatography/mass spectrometry (LC/MS) datasets can be processed in a similar manner to 2-D gel electrophoresis (2-DE) by conversion to images with the mass to charge ratio on one axis and retention time on the other.

  • The conventional image analysis workflow involves detection of protein/peptide spots on each dataset followed by matching of corresponding spots between each dataset and a designated reference.

  • An alternative workflow has emerged, which warps the images into alignment before determining a consensus spot detection on the set of images as a whole.

  • It is important for all researchers to understand the differing unique challenges for analysis tools posed by 2-DE and LC/MS, such as varying operating conditions, artefacts and contaminants.

  • LC/MS analyses also include peptide deisotoping and charge state estimation steps, plus computation of normalised retention times for improved peptide identification.

  • Sound statistical treatment of the results is key, including the use of the false discovery rate correction for overcoming the multiple hypothesis problem.

  • With existing approaches, errors at each stage propagate and amplify as the pipeline progresses; therefore, emerging techniques are aiming to provide an integrated consensus analysis that borrows strength across the image set.

  • Consensus pinnacle detection and image-based alignment and deconvolution are novel methods that can provide automated, reliable and robust quantification.

  • Functional data analysis presents a new statistical paradigm for image-based differential detection able to compensate for multiple experimental factors and discover protein regulation in comigrated regions missed by spot-centric approaches.

Keywords:

  • proteomics;
  • two-dimensional electrophoresis;
  • liquid chromatography;
  • mass spectrometry;
  • image analysis;
  • proteome informatics;
  • spot detection;
  • spot matching;
  • chromatogram alignment;
  • image registration