As part of a continuous work aiming discovery of new prototypes related to natural products
from Brazilian biodiversity, 36 different chalcones were prepared and their trypanocidal activities were
evaluated in vitro [1]. In continuation, this work reports the use of multivariate statistical analysis (MSA)
and machine-learning techniques (MLT) to design new chalcones based on their most efficient groups.
A quantitative structure-activity (QSAR) method was developed with the purpose of predict the EC50,
and CC50 of the designed chalcones. The method used molecular descriptors for the creation of a nonlinear
regression based on artificial neural networks (ANN), which is known for its power for predictions
[2]. Designed chalcones were submitted to the developed method and the biological activity was
predicted against the trypomastigote forms of Trypanosoma cruzi. The experiments were carried out
by creating a dataset containing detailed information about the previously synthesized chalcones and
their correlation with the biological activities. In addition, three-dimensional geometries for each
chalcones were optimized with PM7. The molecular descriptors were calculated and selected via MLT
in Weka 3.8.0. The selected descriptors were submitted to ANN methods with optimized parameters
for the development of a QSAR method. The newly designed chalcones had their geometry optimized
and their molecular descriptors were selected according to the training set. The validation values of
fitness, goodness and predictive power for the prediction models were greater than the minimum
accepted and the scramble method presented very small values, suggesting no overfitting (Table 1)
[3,4]. A clustered heatmap was built with the purpose of classifying the newly designed chalcones
using Euclidean distance in the software MEV 4.9.0 (Figure 1). The QSAR method developed for the
chalcones series resulted in promising values of EC50 and CC50 (Table 2), which classified two of them
in the active group, based on the values of trypanocidal activity reached by the previous chalcones.