The discovery of anti-cancer agents is an area which continues in accelerated expansion. Leukemias (Lkms) are among the most investigated cancers due to its high and dominant prevalence in children. Computer-aided drug design (CADD) methodologies have been extremely important for the discovery of potent anti-Lkms agents, providing essential insights about the molecular patterns which could be involved in the appearance and development of anti-Lkms activity. The present review is focused on the role of the current CADD methodologies for the discovery of anti-Lkms agents with strong emphasis on the in silico prediction of inhibitors against the primary protein associated with the appearance of Lkms: Abelson tyrosine-protein kinase 1 (TPK-ABL1). In order to make a contribution to the field, we also developed a unified ligand-based approach by exploring Quantitative-Structure Activity Relationships (QSAR) studies. Here, we focused on the construction of two multi-targets (mt) QSAR models by employing a large and heterogeneous database of compounds. These models exhibited excellent statistical quality and predictive power to classifying more than 92% of inhibitors/ no inhibitors against seven proteins associated with Lkms, in both training and prediction sets. By using our unified ligand-based approach we identified several fragments as responsible for the anti-Lkms activity through inhibition of proteins, and new molecules were suggested as versatile inhibitors of the seven proteins under study.
Keywords: Artificial neural networks, fragments, inhibitors, in silico design, kinase, leukemias, linear discriminant analysis, molecular docking, QSAR, quantitative contributions