Virtual screening and QSAR formulations for crystal chemistry

In this paper, we demonstrate how partial least squares multivariate analysis can be used to develop a Quantitative Structure-Activity Relationship (QSAR). The analysis is applied to the prediction of bulk modulus based on a combinatorial analysis of crystalchemistry descriptors used in ab-initio calculations. The value of developing these QSAR formulations based on theoretical calculations is discussed in the context of its value as a screening tool for virtual compounds and chemistry.