Bayesian Inference of Regulatory influence on Expression (biRte)

Expression levels of mRNA is regulated by different processes, comprising inhibition or activation by transcription factors (TF) and post-transcriptional degradation by microRNAs (miRNA). biRte (Bayesian Inference of Regulatory influence on Expression (biRte)) uses the regulatory networks of TFs and miRNAs together with mRNA and miRNA expression data to infer the influence of regulators on mRNA expression. Furthermore, biRte allows to consider additional factors such as CNVs. biRte has the possibility to specify Bayesian priors for the activity of each individual regulatory factor. Moreover, interaction terms between regulators can be considered. biRte relies on a Bayesian network model to integrate data sources into a joint likelihood model. In the model mRNA expression levels depend on the activity states of its regulating factors via a sparse Bayesian linear regression using a spikes and slab prior [?]. Moreover, miRNA expression levels depend on miRNA activity states. biRte uses Markov-Chain-Monte-Carlo (MCMC) sampling to infer activity states of regulatory factors. During MCMC, switch moves toggling the state of a regulator between active and inactive and swap moves exchanging the activitiy states of either two miRNAs or two TFs are used [8]. biRte is meant as a replacement for the earlier package birta. biRte offers several advantages compared to birta.