I am trying to find a protocol for performing CD melting experiments on oligonucleotides to analyze their Tm values. All papers mention that they collect the ellipticity at a specific wavelength and at different temperatures, usually between 10-90C with an interval of 0.5-5C. Once the data is collected, plotting the data is done as usual with Sigmaplot or Origin (Temp vs. ellipticity), but the sketchy part is how the curves are fitted. In my case, I usually try and fit Sigmoidal curves using different equations (Gompertz, etc.) and get good R squared values (>0.95), but I understand that R squared values are not the only determinant of a good fit. In some cases I find that the Normality tests and Constant Variance test fail.
There is a very nice nature protocols paper explaining how to fit various models. Especially the Supplemental Information is useful.
Link: http://www.nature.com/nprot/journal/v1/n6/full/nprot.2006.202.html
The easiest, but not perfect way of estimating the quality of your fit is probably a chi-square analysis
Hello Sebastian,
Thank you very much for your reply. I had already referred to that paper but had problems fitting the starting values for the parameters in the equations given in the equations.doc supplementary information file. What would be the starting enthalpy value, and the ellipticity values for the folded and unfolded regions?And how would you set the starting Tm value? The paper mentions that these have to be set prior to calculation, but does not go into detail regarding how to set them for oligonucleotides, and Google does not seem to be my friend when I search for 'starting enthalpy for DNA, etc'. Any help is greatly valued and appreciated!
Regards,
Subramaniyam
What are your expectations from the measurements?
A basic analysis of the nucleic acid melting curve traditionally relies on identifying the maximum of the first derivative of the melting curve to get the midpoint. Even Excel will be able to accomplish this.
If you want thermodynamic parameters, it is more involved. Once you have an equation describing the folded-unfolded equilibrium, you can fit the parameters as a function of temperature using non-linear regression. While this can be done in Excel for simpler equations, it is much better performed using other software. I have most experience with the software package Igor Pro (Wavemetrics), but there are many software packages that allow the fitting of parameters from non-linear regression.
If your system can be described by an equation and would like to forward me the data, I can help to get you started. Reversible two state models are generally not so complex and you should catch on quickly. As you elude to, non-linear regression requires initial "guesses". Here Igor Pro lets you visualise your data and equation using initial parameters to see how close you are.
My experience is with reversible folding and binding of proteins, but the procedure is similar for nucleic acids (but often much more complex - especially if long).
Hello Dr. Berndt,
Thank you for your reply. I am trying to find out the stability of DNA under different environments and as part of my preliminary analysis I am performing CD melting studies.
I am currently using SigmaPlot to do the analysis. My approach has been to take the ellipticity over a range of temperatures (10-90C with a pitch of 0.5C), then normalize the ellipticity between 0-1 using the function
Normalize to [0, 1]:
Y'= (Y-Ymax)/(Ymax-Ymin)
Once normalized, I fit the curve using SigmaPlot, which is usually sigmoidal, and get R squared values of over 0.95. I then plot the equation keeping Y as 0.5 (since the normalization is between 0-1) and get the corresponding X value, which should theoretically indicate the point where 50% of the DNA is in the folded state and the rest is in the unfolded state.
Another approach I have used is the first derivative, where I first normalize the data using the function Y'=Y/Ymax. After plotting, I computed the first derivative and plotted the respective curve. When I tried to fit the curve, I found that no function could satisfactorily fit the derivative curve with good R squared values.
I am extremely grateful for your help and will forward you the data through the ResearchGate messenger. Once again, I thank you for your valuable time.
Regards,
Subramaniyam
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA