Definition of error using LightCycler??? - (Jun/13/2005 )
Hi all!
Reviewers demand a proper definition of ERROR when creating a standard curve using the LightCycler instrument.
What is the definition of "error" in this case, and more importantly; how is it applied? Got a really bad definition from Roche in Sweden that told me nothing, and I can´t find a good definition anywhere on the net or in the users manual. If anyone has a good definition of error in this case, I would be very happy to hear it. I´m not looking for a mathematical definition but for the explanation of what it is and how it can be used.
best wishes
Bempa
Reviewers demand a proper definition of ERROR when creating a standard curve using the LightCycler instrument.
What is the definition of "error" in this case, and more importantly; how is it applied? Got a really bad definition from Roche in Sweden that told me nothing, and I can´t find a good definition anywhere on the net or in the users manual. If anyone has a good definition of error in this case, I would be very happy to hear it. I´m not looking for a mathematical definition but for the explanation of what it is and how it can be used.
best wishes
Bempa
Error= "mean squared error" as calculated by the lightcycler.
The "error" is the amount by which the estimator differs from the quantity to be estimated.
Basically, from what I understood, the mean squared error tells us how much variation exists between your actual data points and the corresponding estimated data points created by the standard curve.
The error gives information on tube to tube variations, probably introduced through bad technique.
Reviewers demand a proper definition of ERROR when creating a standard curve using the LightCycler instrument.
What is the definition of "error" in this case, and more importantly; how is it applied? Got a really bad definition from Roche in Sweden that told me nothing, and I can´t find a good definition anywhere on the net or in the users manual. If anyone has a good definition of error in this case, I would be very happy to hear it. I´m not looking for a mathematical definition but for the explanation of what it is and how it can be used.
best wishes
Bempa
Error= "mean squared error" as calculated by the lightcycler.
The "error" is the amount by which the estimator differs from the quantity to be estimated.
Basically, from what I understood, the mean squared error tells us how much variation exists between your actual data points and the corresponding estimated data points created by the standard curve.
The error gives information on tube to tube variations, probably introduced through bad technique.
Thanks for your answer!
This is the way in which I understood it too. The thing that confuses me is that if you analyse a smaller concentration, then error should be of a greater magnitude than when analysing a high concentration (according to my own experience).
So in my opinion, error should be estimated for each concentration used to generate the standard curve, otherwise it has no real application other than being comparative value between the accuracy of different assays?
Then one of my co-authors asked me if error in this case can be applied to the amplification efficiency (E). Her question was if it is ok to write E+-error when expressing E of an assay. For example E = 1.87 +-0.035. Is this the way in which error is thought to be applied?
best regards
Bempa
"This is the way in which I understood it too. The thing that confuses me is that if you analyse a smaller concentration, then error should be of a greater magnitude than when analysing a high concentration (according to my own experience). "
In total agreement with you.
"So in my opinion, error should be estimated for each concentration used to generate the standard curve, otherwise it has no real application other than being comparative value between the accuracy of different assays?"
The error is basically the sum of the distances between the actual data points and their estimated data points multiplied by 2 and then divided by the total number of data points. So the error is actually sum of the errors generated for each data point. Therefore, the error calculated tells you how reliable your standard curve is.
"Then one of my co-authors asked me if error in this case can be applied to the amplification efficiency (E). Her question was if it is ok to write E+-error when expressing E of an assay. For example E = 1.87 +-0.035. Is this the way in which error is thought to be applied?"
I don't think you can apply the "error" calculation in such narrow fashion. The error can reflect other things besides PCR efficiency variation between tubes.
One example: If one of your standards represents 10000 viral copies/per ml of plasma. You extracted one ml of plasma at that concentration and then elute your extraction in 50 ul. Now you have 10000 copies/ per 50ul of DNA. You use 2ul of your DNA for your reaction. Therefore, for this particular standard, you would have 400 copies of DNA in your reaction. However, in real life pipetting 2 ul will not always yield 400 copies of your virus. Sometimes its a little more or a little less with good technique. However, bad technique (incorrect use of your pipette, inefficient mixing of DNA before use) can lead to greater discrepancies ( 300 or 500 instead of 400 copies going into the reaction). The Error calculation generated is affected by this variance. So while the PCR efficiency's variance tube to tube may be small, the error may be very large because of a mistake of not throughly mixing DNA before use.
PCR efficiency can play a role in "error" calculation. Especially in situation where a someone doesn't throughly mix their master mix before use.
Therefore, the +-error would be part of the whole equation where it could be calculated based on discrepancies of both the starting concentration and/or PCR efficiency.