Elaboração de modelos preditivos da potencial atividade antiviral de espécies da biodiversidade brasileira frente ao SARS-CoV-2 através de métodos computacionais

Abstract

SARS-CoV-2 can cause multisystemic damage with a risk of death and has been responsible for high infection and mortality rates, often underreported. Plant secondary metabolites have shown promise as antiviral agents by strengthening the immune system, protecting against infections, and inhibiting viral replication—particularly through interaction with targets such as the Spike protein and viral proteases 3CLpro and PLpro. Brazil’s vast biodiversity offers great potential for the discovery of new bioactive molecules. Additionally, computational tools such as predictive models and molecular networks accelerate the screening and identification of compounds with activity against COVID-19, optimizing resources and expanding therapeutic possibilities. In this context, the present study aimed to develop predictive models capable of identifying potential plant species and their bioactive compounds by using chemical and biological screenings of extracts from the Fito-Extractoteca, with possible antiviral activity against SARS-CoV-2. The data from these screenings were used to construct predictive models through chemometric methods, as well as to generate molecular networks for qualitative investigation and annotation of the predicted substances. The chemical and biological screening steps yielded promising results in identifying plant species with anti-SARS-CoV-2 inhibitory potential. Moreover, data mining and molecular networking approaches proved to be essential contributors to the refinement of the developed models and to complementary analyses. The results highlight the potential of Brazilian biodiversity as a source of anti-SARS-CoV-2 agents, with multitarget activity through distinct mechanisms. The predictive models optimized the search for anti-SARS-CoV-2 substances and provided meaningful insights for the improvement of subsequent assays. Furthermore, these models—based on experimental screenings— facilitated the study of complex matrices and emerged as promising alternatives to less precise assays, such as molecular docking, or more costly strategies like bioassay-guided fractionation and dereplication. These advances represent significant progress in the search for new drug candidates against COVID-19, by leveraging Brazil’s plant diversity and applying computational methods to streamline the screening and identification of antiviral compounds.