ALDEx2 uses the compositional data analysis methods advocated by Aitchison for over 30 years, coupled with Dirichlet Monte-Carlo sampling to examine high throughput DNA sequencing datasets to identify features that differ between two groups. The only assumption made by the method is that the investigator is sequencing a random sample of the fragments in a given library. This single assumption, coupled with the centred log ratio transformation of Aitchison is sufficient to robustly identify relative abundance differences in datasets as diverse as those from a selective growth experiment, an RNA-seq dataset and a 16S rRNA gene sequencing dataset. It is expected that this approach will be amenable to many other datasets such as ChIP seq and metagenomic analysis