Applied Statistics 2016 /  WS Grzegorczyk

Applying Bayesian networks to reconstruct network structures from data

Marco Grzegorczyk
Johann Bernoulli Institute (JBI), Groningen University, The Netherlands

The reconstruction of regulatory networks from postgenomic data, such as gene expression data, is a very topical research field in systems biology and Bayesian networks are a very powerful tool. Bayesian networks also play an important role in many other research fields, where the goal is to learn the condiditional independency assumptions among variables. Within this workshop a practical introduction to Bayesian network methodology will be given, so that participants not only get familiar with Bayesian networks but also learn how to apply Bayesian networks to reconstruct networks from real-world data. R implementations of all presented algorithms will be provided, and at the end of the workshop the participants will apply the methods to real data and learn how to interpret the inference results. Important topics that are covered within the workshop include: Bayesian networks and the local Markov assumption, directed acyclic graphs (DAGs) and equivalence classes of DAGs, completed partially directed acyclic graphs (CPDAGs), Bayesian Gaussian networks with score equivalence (BGe), Greedy search and Markov Chain Monte Carlo (MCMC) based model inference, and marginal edge posterior probabilities. 

This workshop on Bayesian networks consists of two parts:

The first and main part of the workshop is an introductory (slide-based) tutorial on how to apply Bayesian networks to data.

In the second part of the workshop there will be a practical part where the workshop participants can apply the presented methodologies to infer Bayesian networks from data using the R software. All required R functions will be provided at the workshop.


Workshop participants which would like to participate in the practical part should therefore bring a computer with the R software being installed.

Ideally those participants should also pre-install the two R packages 'limma' and 'ROC'.  Both packages are available from the Bioconductor R package repository.


However, it is important to note that the workshop can also be attended without participating in the practical R programming part.

Dear workshop participants,
here you can download a zip folder with the material (slides and R code) for the AS2016 workshop on Bayesian networks.
Best wishes,