CT-CBN

Continuous Time Conjunctive Bayesian Networks

Current version

0.1.04b Mar 2016

Description

CT-CBN is a collection of two C programs, ct-cbn and h-cbn, that implement algorithms for model selection and maximum likelihood parameter estimation for continuous time conjunctive Bayesian networks. These graphical models have been designed to model the accumulation of mutations subject to constraints on the order in which the mutations can occur. The order constraints are encoded as a partially ordered set (poset) of genetic events. The waiting time for each event follows an independent exponential distribution. 

The original ct-cbn program by Beerenwinkel and Sullivant (2009) implements selection of an optimal poset and estimation of exponential rates from censored data consisting only of the observed mutational patterns. 

The hidden conjunctive Bayesian network model h-cbn accounts for noisy genotype observations (Gerstung et al., 2009). It also implemented a simulated annealing algorithm for structure search, a denoising of the genotypes via the maximum a posteriori (MAP) estimates, and the computation of the most likely progression.

Download

CT-CBN is distributed under the external pageGNU General Public Licensecall_made.

• Downloadct-cbn-0.1.04b Mar 2016 (BZ2, 101 KB)vertical_align_bottom

For changes between versions please see the file CHANGELOG.

Installation

Please consult the README file. The software requires the external pageGSLcall_made library. A compiler with external pageOpenMPcall_made support is highly recommended.

Data sets

We provide a collection of genetic data sets from three cancer studies conducted by the Vogelstein lab. The data consists of three tables with mutations and a mapping of genes to 12 core pathways (external pageParsons et al, 2008call_made). The package also contains the external pagepythoncall_made module cbn.py for parsing the raw data and for bootstrap and permutation analyses. The module requires the external pagenumpycall_made python module.

• DownloadCancer data with core pathways (ZIP, 72 KB)vertical_align_bottom
  

Authors and Contributors

Citations

• Markov models for accumulating mutations
  Niko Beerenwinkel and Seth Sullivant
  external pageBiometrika96(3):645-861call_made

• Quantifying cancer progression with conjunctive Bayesian networks
  Moritz Gerstung, Michael Baudis, Holger Moch and Niko Beerenwinkel
  external pageBioinformatics25(21):2809-2815call_made

• The temporal order of genetic and pathway alterations in tumorigenesis
  Moritz Gerstung, Nicholas Eriksson, C. Jimmy Lin, Bert Vogelstein and Niko Beerenwinkelkel
  external pagePLoS ONE6:e27136.call_made