Inverse analysis of empirical matrices of idiotypic network interactions (Q1361493)

The concept of shape space proposed by \textit{A. S. Perelson} and \textit{G. Oster} [J. Theor. Biol. 81, 645-670 (1979)] has been a useful tool for theoretical immunologists, who have invoked it to model affinity interactions, which play a significant role in mathematical models of immune networks. It has also provided a means to visualize the development of the immune (B-cell) repertoire during the lifetime of a system. The idea of shape space was grounded on the belief from stereo-chemistry and biophysics that binding between protein molecules is effected by generalized shape complementarity. In order to construct a shape space from this approach, one needs to know the significant features of antibody-combining sites necessary for binding to occur. However, in the absence of such specialized information, researchers have had to use hypothetical functional relationships that connect different shapes with the affinities between the molecules they represent. We consider the problem of obtaining shape space - like representations from the reverse direction. Given an empirical matrix of idiotypic interactions (ideally, an affinity matrix), we discuss, with illustrative examples, some methods for graphical summarization of the data. We also discuss the relative merits of this approach versus that proposed by Perelson and Oster, and we suggest possible directions for future research with a view toward applications.