We have developed Discrece Distributional Differential Expression (D3E), a method for identifying differentially expressed genes. Obtaining a list of genes which are significantly different between two conditions (e.g. mutant vs wild-type or stimulated vs unstimulated) is central to many transcriptome analyses. D3E is specifically designed for single-cell RNA-seq data, and it identifies differentially expressed genes by comparing the full distribution of expression levels for the two conditions rather than just the mean expression. Furthermore, D3E is based on an analytically tractable stochastic model of gene expression. The model facilitates interpretation of the results, and it makes it possible to generate hypotheses about the underlying mechanisms behind the change in gene expression.
Download and Installation
D3E is available as a command line tool and can be downloaded from github.
There is also a web-version available for smaller datasets.
The Hemberg group is interested in developing quantitative models of gene expression. Our approach is theoretical and we strive to develop novel mathematical models as well as computational tools that can be used by other researchers.