Extreme learning machines for discovering gene regulatory networks from temporal profiles of expression

M. RUBIOLO; D. MILONE; G. STEGMAYER

Buenos Aires

Conferencia; 4th International Society for Computational Biology Latin America Bioinformatics Conference (ISCB-LA); 2016

International Society for Computational Biology

Reconstructing a gene regulatory network (GRN) from gene expression data is one of the ultimate goals of bioinformatics today. Several computational methods were proposed to support the discovery of GRN, with different levels of accuracy. Most of them require several input sources to provide an acceptable prediction. Indeed, not only gene expression but also other experimental results. We present here a model capable of modeling all possible gene-to-gene regulations just from gene expression data by using a pool Extreme Learning Machine models. High levels of accuracy and specificity were obtained in the reconstruction of three real genomics datasets. Our approach has proven to be accurate and scalable computational tool for the challenging task of identifying the GRN only considering gene expression data.