Computational analysis of single-cell RNA-seq profiles identifies lineage choice and graded transitions in myeloid progenitors

Abstract
Single-cell technologies have gained popularity in developmen- tal biology because they allow resolving potential heterogeneities due to asynchronicity of differentiating cells. Common data analy- sis encompasses normalization, followed by dimension reduction and clustering to identify subgroups. However, in the case of cellu- lar differentiation, we may not expect clear clusters to be present – instead cells tend to follow continuous branching lineages.
We show that modeling the high-dimensional state space as a diffusion process, where cells move to close-by cells with a distance-dependent probability well reflects the differentiating characteristics. Based on the underlying diffusion map transition kernel, we then order cells according to a diffusion pseudo- time (DPT), which allows for a robust identification of branching decisions and corresponding trajectories of single cells. We demon- strate the method on scRNA-seq data of myeloid differentiation. DPT identifies a dominant branching into different myeloid lin- eages and a minor subpopulation of lymphoid outliers. Moreover, a graded transition reflecting erythroid differentiation is identified that dissent from previously stated cluster sequences. We finally identify driver genes and propose how to include additional data sets for integrative analysis across multiple downstream lineages.
Altogether, this illustrates that the concept of discrete transi- tions of progenitors to developed cells may need to be adapted.
http://dx.doi.org/10.1016/j.bdq.2017.02.085