Bayesembler

The Bayesembler is a Bayesian method for doing transcriptome assembly from RNA-seq data.

News

• February 2, 2015: Bayesembler v1.2.0 is released (release notes)
• November 3, 2014: The Bayesembler paper is out in Genome Biology.
• September 4, 2014: The Bayesembler manuscript has been accepted for publication.

In the making

We are currently working on implementing the following features:

• New splice-graph engine (we currently rely on a third party engine) that will support combining replicates for splice-graph assembly
• Single-end read support
• Support for other mappers (e.g. STAR)
• Improved support for unstranded data (assembly accuracy on unstranded data is currently not optimal)
• Improved parallelisation efficiency

Watch us on GitHub or follow us on Twitter (@bayesembler) to stay informed. Please use the GitHub issue tracking system to request any features you would like us to implement.

Installing the Bayesembler

The Bayesembler runs on both Linux and OS X and is freely available under the MIT License. Static Linux x86_64 builds are available under releases (latest release). Please check out the installation-wiki for further information including how to build the Bayesembler from source.

Running the Bayesembler

Input

The Bayesembler requires paired-end RNA-seq data mapped using TopHat2.

Assembly

The Bayesembler can be invoked with default settings using

bayesembler -b <tophat2_map.bam>

Important options include:

• -s: Data is strand-specific. Use -s first to indicate mate orientation as in the dUTP protocol or -s second to indicate opposite orientation. If not provided, the program assumes unstranded data (beware that performance on unstranded data is currently not optimal, we are working on improving this).
• -p: Number of threads used for assembly.

Output

Assembly in Gene Annotation Format (GTF) as assembly.gtf.

Please consult the manual-wiki for additional options and information about the output.

Bug reports, help and feature requests

We want to know if you are experiencing problems with our program or really need us to implement a new feature. To get in touch with us, please use the GitHub issue tracking system or bayesembler@binf.ku.dk.

Citation and credits

The method is described in our paper published in Genome Biology. The Bayesembler is being developed by Lasse Maretty, Jonas Andreas Sibbesen and Anders Krogh at the Bioinformatics Centre, a part of the Section for Computational and RNA Biology at the University of Copenhagen.