Chemometric recovery of latent information from adaptive organosulfur chemical networks

Abstract

Dynamic combinatorial libraries (DCLs) are adaptive chemical systems where reversible reactions generate networks of interconverting species. When multiple exchange reactions share common intermediates, the resulting connectivity allows these systems to retain information about past stimuli in their equilibrium composition. In this work, we show that such hidden information can be quantitatively retrieved through chemometric analysis. A three-layer organosulfur DCL involving thiol/disulfide, thiol/dithioacetal, and thiol/thioester exchanges was exposed to transient basic and thermal perturbations, and the resulting equilibria were analyzed by HPLC-UV. The data were modeled using multiple linear regression (MLR), partial least squares (PLS), and artificial neural networks (ANN). Despite strong collinearity among species, variable selection and dimensionality reduction enabled accurate prediction of the timing and temperature of transient base pulses based solely on equilibrium compositions. Kinetic simulations reproduced the experimental trends and confirmed that the observed chemical memory arises from coupled reversible dynamics. The integration of systems chemistry and chemometrics provides a robust analytical framework for extracting latent information from complex chemical systems.