Detection of differential alternative splicing using Aitchinson's geometry

Gero Doose, Stephan H. Bernhart, Rabea Wagener, Steve Hoffmann


The detection of differentially spliced genes is an important part of the analysis of trancriptomic data. With the growing amount of data available, a fast but sensitive way of detcting significantly differential alternative splicing is needed.

We present DIEGO, a fast but sensitive way of finding alternative splicing events in expression data based on either exon or splice-junction usage. DIEGO is orders of magnitude faster than competing programs such as DEXSeq, and can cope with group sizes that are not feasible with other programs.

DIEGO is based on Aitchinson's geometry, as is fit for composite data. Differential alternative splicing can be found irrespective of expression differences and, using mapped splice junction information e.g. from segemehl, tophat2 or STAR aligners, also is independent from isoform knowledge. DIEGO is implemented in python and comes with no warranty.

Link to the publication.


The software is published under the GNU GPL v2.0 license.