;)
Does a low mean CV equal reproducibility?
Posted by admin in 2DE Knowledge Base
 Johan Ljunggren, Key Account Manager at Ludesi answers |
Q: “Can I use the mean CV to measure the reproducibility of 2D gel image analysis data?” |
The CV is the standard deviation divided by the mean. In this context it is the standard deviation of the normalized spot volumes of a specific protein or PTM, divided by the mean of these spot volumes. The mean CV is the average CV of all the proteins in an experiment.
The lower the mean CV the less variation is present between spot volumes in the different gels, meaning that the reproducibility is higher. The question is however how these volumes are calculated.
Let’s say for instance that a particular image analysis software generates large spot borders that encompass gel material outside the actual protein spot and sometimes even detects several distinct spots as a single spot. In this case, the calculated volumes will generally be larger and the standard deviations smaller, and CVs will consequently be lower.
Or let’s say the particular analysis software subtracts a relatively low amount of background in the course of calculating the spot volumes (or the user chooses to not subtract any background at all, as is possible in certain software packages). This will also lead to larger spot volumes and hence to lower CVs.
In such cases, the image analysis might lead the researcher to believe the variation between the gels is low as measured by the mean CV, when in reality this is merely an effect of the image analysis method.
Therefore, if the mean CV is used to indicate the reproducibility of a 2D gel set, the result can only be compared to the mean CV of gel sets analyzed with the same software, using the same settings. Otherwise, the disparity in mean CV may just as well be due to differences in the image analysis as to differences between the gels themselves.
An alternative to using mean CV to measure variation would be to use the average Spearman rank correlation coefficient between all gel pairs, because the Spearman correlation does not suffer from the problem of being affected by an overall increase in spot volumes.
;)
