Global fitting: the key for a robust analysis

Download use case: Global Fitting


The Indian parable of “The six blind men and the elephant” tells the story of six blind men who touch an elephant in the hope of learning what it is like. As each one can only feel a different part of the animal the individual conclusions obtained are in disagreement and none of them provides a real view of the full elephant. “only by sharing what each of you knows can you possibly reach a true understanding”; that´s the moral behind this nice story.


Fig 1: The six blind men and the elephant: only a global analysis of the overall data provides a true understanding.

The binding assay(s) achieved to characterize a molecular interaction often provides not just one, but several binding curves from which the affinity constant is obtained.
Sometimes, an individual fit of these curves yield a set of binding constants that are significantly different from them; this result can be very confusing because, in principle, these binding curves are a representation of the same binding event and should converge to provide the analogous information. Often, the explanation for this behaviour is that the different curves indeed provide only partial and/or different information of the interaction, not enough to unequivocally determine the binding affinity through individual analysis.

“It´s like feeling only a separate part of the elephant”

This is a typical scenario when facing the study of complex binding events that involve more than one equilibrium and several binding curves are obtained, i.e. from different frequencies of the spectra in a titration experiment, from data registered using different techniques (ITC, NMR, Optical Spectroscopies…) and/or from experiments performed at different concentrations of the species participating in the binding event.

Analogous to the parable of the six men and the elephant, the way to get a true understanding of the binding event consist of the global analysis of the different curves.

Fig 2: The binding curve obtained from 2D NMR titrations.

Being aware of the relevance of global analysis, in AFFINImeter we count with the possibility to perform Global fitting of multiple data to tailored binding models where one or more fitting parameters are shared between isotherms. The number and identity of the parameters shared are selected by the user.
Moreover, two or more parameters can be related through mathematical relationships designed by the user. All these features make our global fitting tool the most potent among others to perform a robust analysis of binding data of complex interactions.

How to get the most out of biophysical techniques to address binding interactions?

The analysis of isotherms is the more direct way to calculate binding constants for molecular interactions. Isotherms can be obtained using different techniques (ITC, SPR, NMR, Uv-vis, IR, Fluorescence, Circular Dichroism…) and at different experimental conditions.


1. The global fitting approach allows to simultaneously analyze several isotherms obtained by different biophysical techniques and/or at different experimental conditions in a very accurate manner.



2. Many interacting systems do not bind with a simple 1:1 model, more complex binding model can be designed to address complex interaction

3. Using tools to globally analyze isotherms obtained by different biophysical techniques is the most reliable method to characterize binding interactions by the orthogonal approach.

Click here to star using AFFINImeter and to start to create your own binding model:



AFFINImeter binding models for Nuclear Magnetic Resonance

AFFINImeter is already well known in ITC binding data analysis for providing the possibility to use tailored binding models created by the user. The models are generated with the tool “model builder” that includes a letter code “M-A-B” to describe titrate (M), titrant (A) and if necessary, the presence of a third species (B) (Figure 1).


Fig.1 Example of a competitive binding model created in AFFINImeter where the titrant in syringe “A” binds to the titrate in cell “M” to form a 1:1 complex “MA” and a second ligand “B” mixed in the cell with “M” forms the complex “MB” and thus competes with “A”.


Following the same approach, the binding models available for the software AFFINImeter for Nuclear Magnetic Resonance are generated with the model builder and based on the “MAB” code. But there are significant differences between ITC and NMR data analysis when the time comes to select a binding model from AFFINImeter, which have an origin in the inherent characteristics of each technique and in the different experimental design. In chemical shift perturbation (CSP) NMR titration experiments, the observed parameter used to monitor the progress of the binding event is the chemical shift of titrate resonance signals. Hence, the models used for NMR data analysis require the presence of compound “M” (titrate) as it is the species from which changes associated with the binding process are monitored. Conversely, in ITC the observed parameter is the heat change upon interaction and this parameter is not necessarily linked to a particular species “M”, “A” or “B”.

An illustrative example is the evaluation of a monomer-dimer self-association process using NMR or ITC. In NMR, the standard experimental setup would consist in the incremental dilution of the compound sample at high concentration in the NMR tube, to monitor dimer dissociation (Figure 2a). In ITC the standard experimental setup would consist in a titration of the compound sample at high concentration in the syringe (species “A” according to the AFFINImeter code) into the calorimetric cell filled up with solvent (Figure 2b).


Fig.2 Representation of experimental setup for a) NMR dilution experiment and b) ITC dilution experiments. The corresponding schemes of AFFINImeter binding models for data analysis are shown.



Would you like to know more about AFFINImeter for Nuclear Magnetic Resonance? Press the button below:



AFFINImeter new version!

During the last months, we’ve contacted you asking your opinion and experience with the software. Thanks for all your suggestions and needs we have improved the previous version of the software to make it easier, faster and more versatile.

What is new in this release?

  • In collaboration with Mestrelab Software, AFFINImeter can now analyze Nuclear Magnetic Resonance (NMR)  titration curves.

  • The Main toolbar has been simplified:

Data” & “Projects” have been merged into a single “Projects & Data*” menu

You can filter the elements to show by selecting “Projects & Data” (default), only “Projects” or only “Data“.

A new option has been included to easily pre-visualize the settings of your experiments directly from the list of data & projects.

The “Instruments” management menu has been removed.

*Data elements are represented by this icon –>  
*Projects elements are represented by this icon–> 


  • Now you can organize all your data and projects into folders:


  • Organize your data & projects as you would do on your computer.
  • Looking for your already existing KinITC data? Go to Projects & Data, and look into the folder “KINITC”.
  • A new parameter has been introduced to measure the quality of the FIT: Goodness of Fit

The GoF (Goodness of Fit) is the probability of finding the fitted value within a normal distribution with half-width equal to the uncertainty of each experimental point. The GoF for a curve is obtained as the mean GoF of all its points. The ideal value of perfect fit would be Gof=100%.

This parameter will be available when you execute a new FIT project/or re-execute an older project. If you visualize an already executed project, the previous parameter (χ 2) still will be present.

What have we improved from the previous version?

  • We have incorporated the possibility to perform blank subtraction right after raw data upload and processing.
  • Multiple minor usability improvements and performance tweaks.
  • Multiple bug fixes and issues suggested by our users have been implemented and resolved.


Try the new AFFINImeter here