turinglife / bnsa Goto Github PK

We present a computational tool, BNSA, Bayesian Network with the Splitting-Averaging strategy for discovering gene regulatory relationships.

BNSA utilises Bayesian network structure learning and extends it by considering different sample conditions in order to effectively identify both strong and subtle interactions from gene expression profiles. The tool makes use of heterogeneous data, including putative target Information, gene expression profiles of both gene regulators (e.g. microRNAs) and mRNAs, and sample categories.

We have applied BNSA to discover microRNA (miRNA) regulatory relationships in the data sets of EMT (epithelial-to-mesenchymal transition). The result shows that BNSA discovers more biologically relevant miRNA-mRNA interactions compared to normal Bayesian Networks. We have also used BNSA to learn from heterogeneous data the three-component regulatory networks, with the presence of miRNAs, Transcription Factors (TFs) and mRNAs. The findings elucidate the complex gene regulatory mechanism which involves both TFs and miRNAs. Several discovered interactions and molecular functions have been confirmed by literature.

BNSA was initially coded in MATLAB. An R package for BNSA is currently being developed and it will be made available publicly in near future.