ABSTRACT
A mathematical procedure is proposed for the analysis of multivariate data recorded during spectroscopically monitored melting experiments of biomolecules such as nucleic acids and proteins. The method is based on hard/soft hybrid modeling in which one part of the observed variance is explained in terms of a physicochemical model (hard modeling), whereas the other part of the observed variance is explained in terms of soft modeling. The physicochemical model is applied to all of the components related to the unfolding of the biomolecules studied and provides thermodynamic values associated with the unfolding process such as the change in enthalpy, entropy, and melting temperature. The soft modeling term explains the contribution of artifacts not related to the unfolding process such as baseline drifts and nonlinearities. Here the method is applied to the analysis of simulated and experimental data corresponding to the unfolding equilibria of intramolecular structures such as i-motif and G-quadruplex. Overall, the method provides better results than the commonly used univariate approach and also better results than pure hard modeling.