How does it work?
When a map is created, a mesh of 720 different prototype 1:1 interaction is created. These correspond to one pixel each in the map.
Next, a non-linear fitting algorithm adjusts the contribution of each of these 720 to make the sum of all prototype interactions fit the measured curve(s).
A constraint is added to promote smooth surfaces, this is called regularization. In brief, if the fitting algorithm can choose between a rough surface and
a smooth surface that gives equally good fits, regularization will guide the fitting algorithm towards smoothness leading to a minimum number of peaks in the map.
It is important to remember that the Interaction Map method cannot differ between a true binding event and other biological events that have a
similar temporal development as molecular interactions. It is for example possible that the internalization of a protein bound to a surface receptor of a
cell would appear as a peak in the map. As always, relevant control experiments are required to pinpoint the nature of each peak in the map.
It is also important to emphasize that Interaction Map does not in itself lead to any structural understanding.