;)
The dangers of using a reference gel when matching gel images.
Posted by ludesi in 2DE Knowledge Base
 Johan Ljunggren, Key Account Manager at Ludesi answers |
Q: “What is the problem of using a reference gel when I am warping (matching) my gels?” |
The use of a master gel has the downside that certain spots may not be present in the master gel, and that consequently these spots cannot be matched. This problem becomes more pronounced the greater the difference is between the gels. A possible solution is to run a gel with a pool of samples from the study, and use this gel as the master gel. This gel will be more likely to contain most or all the spots present in the other gels. However, doing so takes additional time, effort and money.
Another problem with the use of a master gel is that there will be information present in the gel set that could have been used to optimize the matching that is in fact not being used. With a master gel, only information relating to the relation between the master gel and the rest of the individual gels is taken into account, while the great body of information pertaining to the relation between all gels in the study is not being put to use.
Ludesi handles matching by performing what we have termed an all-to-all matching. This matching employs an algorithm that attempts to make the best possible matching decisions based on information from the relations between all gels simultaneously without the use of a master gel, resulting in improved matching performance.
Yet another issue with the use of a master gel is that it may cause an artificial extra degree of similarity to the group of samples to which the master gel belongs. Consider the case of an experiment with three different groups of gels: A, B and C, and one of the gels in group A is selected to be the master gel. This setup is likely to cause an extra degree of similarity between groups A and B, and between groups A and C. Consequently, groups B and C will falsely appear less similar, relatively speaking. This issue is not merely theoretical but is something we have encountered ourselves in actual analysis data generated by software that makes use of a master gel.
Warping is the process of distorting gel images so that the spot patterns become aligned when the images are placed on top of each other, whereby it can be inferred that spots at the same location in different images belong to the same protein or PTM.
If warping is performed by warping all gels to a single master gel, then this warping suffers from problems similar to those associated with matching to a master gel. The master gel may not contain all spots which makes the warping process more difficult and prone to error. The analysis may also show a false degree of extra similarity to the group of gels the master gel belongs to.
Some software packages include the option of selecting which gels are to be warped to one another instead of forcing the warping of all gels to a single master gel. Using this method, the most similar gels may be warped to one another, provided there is a connection in the chain of warping between all gels in the experiment. In this fashion, some of the problems associated with the use of a master gel may be avoided.
;)
