Entropy measures are universal parameters useful to codify biologically-relevant information in many systems.
In our previous work, (Gonzalez-Diaz, H., et al. Chem. Res. Toxicol. 2003, 16, 1318–1327), we introduced the molecular
structure information indices called 3D-Markovian electronic delocalization entropies (3D-MEDNEs) to study the quantitative
structure-toxicity relationships (QSTR) of drugs. In a second part, (Cruz-Monteagudo, M. et al. Chem. Res. Toxicol.,
2008, 21 (3), 619–632), we extended 3D-MEDNEs to numerically encode toxicologically-relevant information present
in Mass Spectra of the serum proteome. These works demonstrated that the idea behind classic drug QSTR models
can be extended to solve more general problems in toxicological chemical research. For instance, there are not many reports
of multi-target QSTR (mt-QSTR) models useful to predict multiplexed endpoints of drugs in a high number of cytotoxicity
assays. In this work, we train and validate for the first time a QSTR model that correctly classifies 8,806 out of
9,001 (Accuracy = 91.1%) multiplexing assay endpoints of 7903 drugs (including both training and validation series).
Each endpoint corresponds to one out of 1443 assays, 32 molecular and cellular targets, 46 standard type measures, in two
possible organisms (human and mouse). We have also determined experimentally, for the first time, the values of EC50 =
8.21 μg/mL and Cytotoxicity = 26.25 % for the antimicrobial / antiparasitic drug G1 on Balb/C mouse thymic macrophages
using flow cytometry. In addition, we have used the new model to predict G1 endpoints in 1,283 assays finding a
low average probability of p(1) = 0.50% in 152 cytotoxicity assays. Last, we have used the model to predict average probability
of the interaction of G1 with different proteins in macrophages. Interestingly, the Macrophage colony-stimulating
factor receptor, the Macrophage colony-stimulating factor 1 receptor, the Macrophage migration inhibitory factor, Macrophage
scavenger receptor types I and II, and the Macrophage-stimulating protein receptor, have also very low average
predicted probabilities of p(1) = 0.092, 0.038, 0.077, 0.026, 0.2, 0.106, respectively. Both experimental and theoretical results
show a moderate thymic macrophage cytotoxicity of G1. The obtained results are significant because they complement
the immunotoxicology studies of this important drug.