| Summary: | While progresses have been made in mapping transcriptional regulatory networks, posttranscriptional regulatory roles just begin to be uncovered, which has recently arrested much attention due to the discovery of miRNAs. Here we demonstrated a combinatorial approach to incorporate transcriptional and posttranscriptional regulatory sequences with gene expression profiles to determine their probabilistic dependences. Both miRNA target motifs (miRNA-mediated posttranscriptional regulatory sites) and TFBSs (transcription factor binding motifs) were incorporated with microarray time course gene expression profiles to determine the network of dependencies among factors involved. Bayesian network model was introduced to deduce the conditional dependences between expression profiles and the combinations of two types of sequence motifs. Based solely on the sequence motifs adopted in the network models, we could correctly predict expression patterns for more than 50% of 1,132 genes, which was statistically significant. The result suggested that microarray time course dataset could be used to detect the change of mRNA steady-state level which might be affected by miRNA regulation, and the combinatorial approach was efficient to determine the underlying context-dependent roles in transcriptional and posttranscriptional regulation. The computer programs were written in C++ and PERL to implement the proposed methods.
|
|---|