Other approaches to quality control of 2D gel image data

Ad hoc methods used in various articles

Previous research has often involved comparisons of the performance of different software packages for 2D gel image analysis. Other studies have attempted to construct artificial gel images as input for the assessment of the performance of image analysis software.

In our opinion, there has been a lack of focus on the measurement of correctness itself, on real gel images, and different authors have used their own ad-hoc methodology to evaluate various aspects of image analysis. This makes comparisons between studies difficult.

Reproducibility vs correctness

Today there is a great deal of focus in the proteomics community on reproducibility, but reproducibility should not be confused with correctness. The concept of reproducibility does not take into account whether or not the data is correct, only if it is similar. A narrow focus on reproducibility is insufficient, because data may well be reproducible, but incorrect.

Reproducibility only makes sense if the reproducible data is correct. Correct data that can not be reproduced give valuable (if yet negative) feedback about the experiment. But reproducible data that is incorrect can be seriously misleading if the reproducibility of the data is used as an argument that it is trustable. In short, reproducibility without correctness has little if any value to researchers.