The article, Understanding melting temperature (Tm), describes considerations for better oligonucleotide design, explained by IDT biophysics scientist and resident Tm expert, Dr Richard Owczarzy. In that article, Dr Owczarzy provides a formula for accurate Tm calculation. Here, he responds to a researcher who recently asked us if it was critical to use nearest neighbor parameters for calculating the most accurate Tm. The researcher noted that such Tm values are almost always higher, and sometimes substantially higher, than Tm values calculated using other formulas.
Key observations
It is not sufficient to use the nearest-neighbor parameters to achieve the most accurate Tm predictions. The program must also take into account PCR conditions and oligo concentrations (for a discussion, see the article, Understanding melting temperature (Tm)). PCR conditions can vary, so ideally, a user should enter the exact reaction conditions into software. The IDT OligoAnalyzer® Tool allows you to do that.
Oligonucleotide melting temperature is not constant, but changes significantly with duplex environment. Nearly all nearest-neighbor parameters have been measured in 1 M Na+ buffer, which is far from standard biological buffer concentrations. Therefore, programs must be able to make corrections for your specific PCR conditions, that is, where Tm is lowered due to lower salt concentration. The algorithm needs to take into account both monovalent (Na+, K+) and divalent (Mg2+) cations. The OligoAnalyzer tool considers all of these issues and applies non-linear salt correction that is more accurate than the linear versions used in some programs [1].
Some old programs may also use outdated nearest-neighbor parameters; e.g., Breslauer et al. (1986) [2]. Tm predictions based on old parameters were shown to have up to 3X greater average error than the recent unified parameters [3,4]. This is not surprising because the latter parameters were derived from large data sets. This is an active area of research and similar updates of thermodynamic parameters have occurred for predictions of RNA-RNA duplexes.
Other simple formulas, such as methods based on GC composition (GC%), often assume an average PCR buffer, and average primer length and concentration. But most experimental conditions are not average. The simple formulas do not accurately model the effects of base sequence, sequence length, and probe concentration. Tm values derived from these formulas can easily have substantial errors (±10°C) if your experimental conditions are not average; e.g., primer GC% deviates from 50% or Mg2+ concentration deviates from the average 2 mM.
Recommendations
Many software tools and apps predict duplex stability. Accuracy varies widely. It is therefore a good idea to review the descriptions of the calculation formulas, parameters, and options. Accuracy of predictions is likely to be low if the software tool does not apply the latest nearest-neighbor parameters or does not correct for the effects of salt and oligo concentrations. The articles cited under References report experimental methods for measuring Tm and are a useful resource for evaluating software accuracy.