Hi!

I would like to collect your opinion about estimating efficiency of qPCR reactions from the amplification curve using a range of fluorescence readings within the log-linear phase (like the LINREG procedure does).

I have seen several papers describing the method, it sounds very logical to me, and I personally find it more realistic than the dilution series and less prone to uncontrollable error. I have not been able to find critics of the method, so I wanted to know basically if you could offer arguments against it or if you know of any paper in which they critically analyse the method.

Thanks!

Submit your paper to J Biol Methods today!

# Efficiency estimation using a window of linearity

Started by echica, Apr 13 2010 06:14 AM

3 replies to this topic

### #1

Posted 13 April 2010 - 06:14 AM

### #2

Posted 13 April 2010 - 09:07 AM

I tried few times to estimate efficiency from the reaction curve with the PCR Miner application, but the efficiencies were always quite different from those measured by dilution series (usualy lower) and I wasn't sure which one was the more acurate one.

So I sticked with using the dilution method, because that's the standard used in many papers today. And because it's more convinient, the instrument program computes the efficiency for me.

So I sticked with using the dilution method, because that's the standard used in many papers today. And because it's more convinient, the instrument program computes the efficiency for me.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.*'Normal' is a dryer setting.* - Elizabeth Moon

### #3

Posted 13 April 2010 - 10:15 AM

I have tried linear regression of efficiency (LRE) - http://www.biomedcen.../1472-6750/8/47 many times with no luck. The efficiency was always very different and not good. I managed to get log-linear phase into linear detection range of Stratagene Mx3005P but it did not help anyway. SYTO-9 instead of SybrGreen did not help too. I gave up after five 96-well plates. However I still think that this would be the best way to assess efficiency because every template is unique and can affect PCR efficiency differently.

What I found is:

1. you have to optimize machine detector gain setting or sample volume or tube plastic type in order to measure S phase in linear detection range

2. you have to optimize primer concentrations to get absolute clean single-peak product

3. do at least 3 replicates

I do not know why it did not work well for me however you might have more luck with this method.

What I found is:

1. you have to optimize machine detector gain setting or sample volume or tube plastic type in order to measure S phase in linear detection range

2. you have to optimize primer concentrations to get absolute clean single-peak product

3. do at least 3 replicates

I do not know why it did not work well for me however you might have more luck with this method.

### #4

Posted 15 April 2010 - 12:43 PM

Thank you very much for the replies, and sorry for my late reply.

I have tried a couple of times to extract the efficiency from the amplification curves when I ran some dilution series, and I also found that efficiencies calculated from regression were lower than the ones calculated from the dilution series. However, at that time I felt more confident about using them as I was able to "see" that the increment in fluorescence from cycle to cycle were more consistent with the efficiency calculated from the regression than from the dilution curve (i.e. the dilution curve gave nice efficiency but the actual change in fluorescence was not as good). This made me curious about this method.

Then I had my crisis of faith. I agree with Trof in that the dilution series is the preferred method (sometimes even required) by some journals and review panels to report efficiency in qPCR, and this is exactly what puzzles me. I have not been able to find reasons why efficiencies calculated from dilution series are considered more "reliable" than the efficiencies from the regression methods (here I assume journals and reviewers prefer method because is more reliable). On the other hand, I have not been able to find evidence against the regression method for calculating efficiencies.

I was able to find some arguments that I consider valid against efficiencies calculated by fitting a non-linear regression on the whole amplification curve, as the models used for regression are not accurate in all the phases of the amplification curve (they normally are accurate in phase but not the next). However, if only the log-linear phase is used (exactly around where we estimate the Cq values), then linear regression does a pretty good job, and agreeing with vladoo, doing this or something similar to this would be more reliable as reaction efficiencies are pretty much reaction dependent.

I wonder whether there will be some major methodological shift in calculating qPCR efficiencies in the near future, I have started to seen a good number of papers in which they report having used the regression methods to check their qPCR efficiency. I guess for now that safest thing could be to have a good dilution series at hand, anyway, efficiency from the regression could be calculated at any time if you keep your raw fluorescence data, but the same is not true for dilutions.

Thank you very much again for your input, I appreciate very much your thoughts.

I have tried a couple of times to extract the efficiency from the amplification curves when I ran some dilution series, and I also found that efficiencies calculated from regression were lower than the ones calculated from the dilution series. However, at that time I felt more confident about using them as I was able to "see" that the increment in fluorescence from cycle to cycle were more consistent with the efficiency calculated from the regression than from the dilution curve (i.e. the dilution curve gave nice efficiency but the actual change in fluorescence was not as good). This made me curious about this method.

Then I had my crisis of faith. I agree with Trof in that the dilution series is the preferred method (sometimes even required) by some journals and review panels to report efficiency in qPCR, and this is exactly what puzzles me. I have not been able to find reasons why efficiencies calculated from dilution series are considered more "reliable" than the efficiencies from the regression methods (here I assume journals and reviewers prefer method because is more reliable). On the other hand, I have not been able to find evidence against the regression method for calculating efficiencies.

I was able to find some arguments that I consider valid against efficiencies calculated by fitting a non-linear regression on the whole amplification curve, as the models used for regression are not accurate in all the phases of the amplification curve (they normally are accurate in phase but not the next). However, if only the log-linear phase is used (exactly around where we estimate the Cq values), then linear regression does a pretty good job, and agreeing with vladoo, doing this or something similar to this would be more reliable as reaction efficiencies are pretty much reaction dependent.

I wonder whether there will be some major methodological shift in calculating qPCR efficiencies in the near future, I have started to seen a good number of papers in which they report having used the regression methods to check their qPCR efficiency. I guess for now that safest thing could be to have a good dilution series at hand, anyway, efficiency from the regression could be calculated at any time if you keep your raw fluorescence data, but the same is not true for dilutions.

Thank you very much again for your input, I appreciate very much your thoughts.