Estudo da entropia de tsallis para a inferência de redes gênicas

Abstract

The amount of information in a system can be measured by entropy. A particular case of a system is a network formed by the interaction between genes, known as gene networks. In this work we study how one type of non-extensive entropy, Tsallis entropy, can provide the greatest amount of information for gene networks, through the choice of the best non-extensive parameter q. It is shown that it is possible to obtain numerically the best parameter, and that it depends on the number of degrees of freedom of the system, in the binary case the best value being approximately 2.46. This result is tested in the context of gene network inferences, initially with logic gates, followed by artificial gene networks and finally with experimental data obtained from the DREAM4 challenge. At last, these results are compared with results from previous works, indicating the adequacy of Tsallis entropy for the inference of gene networks.