Reconstruction of the antibody affinity distribution from experimental binding data by a minimum cross-entropy procedure.

A new solution is presented for the reconstruction of the distribution of association constants of antigen-antibody binding from a finite number of noisy experimental binding data. This ill-posed problem is solved by utilizing an information-theoretic method based on the principle of minimum cross-entropy (MCE) to select, as the solution, that unique antibody binding distribution which minimizes the cross entropy relative to some prior distribution subject to the constraints imposed by the given measurements. The prior distribution is selected to properly encode all the a priori information on the affinity distribution before the measurement of the experimental binding data. The utility of the method is demonstrated by application to synthetic binding data.