Protein Thermal Shift™ 软件 v1.4

We recommend looking at the spread in Tm, which is more important than the relative fluorescence.

Some proteins have hydrophobic residues on the surface and the dye binds to these residues. Heating results in unfolding of the protein causing more hydrophobic residues to be exposed. The dyes bind preferentially to these inner locations and so there is a flattening (or a very low rise) of the melt discernable in the melt profile. If there is no positive slope, you will not get a Boltzmann Tm, but you should still get a derivative one. And you can always draw a manual region to get a Tm out. Some proteins will not work with this technology if the hydrophobic residues are already exposed on the surface and the dye binds strongly to it. Please contact Technical Support at techsupport@thermofisher.com about the possibility of other dyes being available for this issue.

Make sure that you first open the file in the corresponding instrument software, click “Analyze”, and then save the file, before trying to open with the Protein Thermal Shift Software. The file must first be analyzed before it can be used in the Protein Thermal Shift Software.

The software will allow for ≥100 plates per study. We allow the user this flexibility but do not recommend you mix data from multiple plates unless they have validated their results in advance. At a minimum, we recommend researchers include a reference assay in each plate to ensure reproducibility.

We provide two independent methods because they each have unique things to offer in terms of the analysis. The two-state Boltzmann model has a physical meaning and appeal. It also provides a great way to normalize across noisy undulations in the signal. However, those undulations may be of actual interest and not noise, such as for multi-domain proteins where they may correspond to different domains coming apart in stages. Here the two-state model is inappropriate. The derivative method can help get a temperature at which the local peaks occur. These are two completely unrelated approaches. If the two-state model is a great fit for your data, the results should be in close agreement.