Our research focuses on the development of models, algorithms and visualizations that compose computational strategies to deal with essentially biological problems. We use graph models and complex network concepts, as well as machine learning and data mining to understand macromolecules, especially the 3D structures of proteins, their interactions and functions.
We seek to understand the recognition process between proteins and ligands through computational strategies at the atomic level. Such a process is an important step towards predicting ligands, identifying targets, lead discovery and rational design of molecules. We are interested in ligands that are small non-protein molecules and also in ligands that are peptides or proteins. We have developed computational tools that detect conserved substructures at the protein-ligand interface. This allows us to calculate representative structural motifs of the interaction interface for each protein family. These motifs, together with our studies to predict the impact of amino acid mutation on protein structures, can be used as input for the prediction/design of molecules to interact with a target of interest.
Considering that our group is in the context of the Universidade Federal de Viçosa, recognized for its agrarian vocation, we have also been working with the proposal of computational strategies of artificial intelligence to deal with problems related to agriculture, such as those involving the characterization and prediction of peptides from plants with antimicrobial activity, prediction of protein abundance using codon usage bias of genes, and plant-pest interactions.
We work with visualization in our bioinformatics studies to present the results so that trends, patterns, exceptions and anomalies can emerge, supporting domain specialists in understanding these results. We use data visualizations to help us make our machine learning strategies more explainable, so that not only the results are presented, but domain experts can gain insights into the process of how they were calculated and to what extent the descriptors impact on such results.
Public sector projects
We have been working on a research project entitled "Observatory of Public Policies for Agriculture and Rural Areas" with the Brazilian Ministry of Agriculture using visualization strategies to transform data into information. From a large volume of data, containing the temporal dimension, our visualization dashboards provide timely and understandable assessments to support decision making. The project aims to analyze historical data to support the creation and evaluation of public policies for agriculture. This project is developed in the context of the Institute of Public Policies and Sustainable Development (IPPDS) at UFV, under the coordination of Professor Marcelo Braga. Professor Silveira coordinates the software development team for this project.
SANTANA, Charles A. et al. GRaSP-web: a machine learning strategy to predict binding sites based on residue neighborhood graphs. Nucleic Acids Research, v. 50, n. W1, p. W392-W397, 2022.
SANTANA, Charles A. et al. GRaSP: a graph-based residue neighborhood strategy to predict binding sites. Bioinformatics, v. 36, n. Supplement_2, p. i726-i734, 2020.
PAIVA, Vinicius de A. et al. Protein structural bioinformatics: An overview. Computers in Biology and Medicine, p. 105695, 2022.
RIBEIRO, Vagner S. et al. visGReMLIN: graph mining-based detection and visualization of conserved motifs at 3D protein-ligand interface at the atomic level. BMC bioinformatics, v. 21, n. 2, p. 1-12, 2020.
QUEIROZ, Felippe C. et al. ppiGReMLIN: a graph mining based detection of conserved structural arrangements in protein-protein interfaces. BMC bioinformatics, v. 21, n. 1, p. 1-25, 2020.