Métodos de inferência de redes gênicas: uma revisão avaliativa do estado da arte

Abstract

Gene networks are representations of the interactions between genes and are very important for understanding gene interactions holistically and thus estimating their functioning in response to a certain stimulus. Therefore, the process of inferring gene regulatory networks (GRNS) is extremely significant in the field of bioinformatics to understand the mechanisms that govern interactions and regulations between genes during biological processes. At the same time, due to great advancements in the area, new inference methods are frequently developed using diverse computational and mathematical approaches. Thus, it is important to compare these techniques in order to verify if they have common responses, which could increase the accuracy of the obtained results. In this context, this work proposes to carry out an evaluative review of infe- rence methods considering current approaches that operate on both bulk and single-cell data. The dataset adopted for the evaluation, provided by the DREAM5 repository, was based on the bacterium E. coli. The objective of this research is to make a significant contribution to the field of computational biology, providing a basis for the development of more efficient and ac- curate methods. The results from the intersection analysis reveal that certain methods, such as EnTSSR, IEntropy, SCENIC+, and SPIDE, show significant correspondences, indicating potential complementarity. This correlation suggests that the combined use of methods can provide more robust and reliable results than their isolated use, pointing to a promising strategy in gene network inference.