![]() First, each dimensionality reduction technique has a specific bias that determines which type of information is preserved in the reduction. Current approaches visualize the cell types using dimensionality reduction techniques like principal component analysis (PCA), multi dimensional scaling (MDS) or t-distributed stochastic neighbor embedding (t-SNE) 16, which allow the easy detection of instances (cells) that are distant from cluster centers, thus pointing to possible differentiation pathways. The million dollar question therefore is how to integrate both views in the most efficient way. In particular, clusters frequently represent metastable intermediate differentiation stages or stable end points, respectively, and can thus serve as anchor points, facilitating the derivation of differentiation trajectories. However, it would be much more useful to combine clustering with differentiation pathway visualization since the clustering of major cell types can serve as an excellent validation tool. Examples are Monocle 13, which determines a pseudo-time associated with differentiation progress from the similarities between cell profiles, the use of diffusion maps to directly determine differentiation trajectories 14, or graph-based approaches like Wishbone 15. One possible solution is to give up at the detection of subpopulations and cell identities altogether. This is especially true for rare cell types such as stem cells. While this line of research is very successful in determining main cell types, the clustering hypothesis implies a discretization that does not reflect the nature of differentiation as a continuous process. The development of tailored clustering approaches, including measurements for the similarity of transcriptome profiles, is complex and subject to active research 4, 7, 8, 9, 10, 11, 12. ![]() 1, 2, 3, 4, see 5, 6 for recent reviews). State-of-the-art approaches for cell type classification use clustering to identify subpopulations of cells that share similar transcriptional profiles (e.g. One of the most important tasks in single-cell RNA-seq is to identify cell types and functions from the generated transcriptome profiles. We show on intestinal epithelial cells and myeloid progenitor data that GraphDDP allows the identification of differentiation pathways that cannot be easily detected by other approaches. ![]() GraphDDP starts from a user-defined cluster assignment and then uses a force-based graph layout approach on two types of carefully constructed edges: one emphasizing cluster membership, the other, based on density gradients, emphasizing differentiation trajectories. Here we present GraphDDP, which integrates both viewpoints in an intuitive visualization. Crucially, clusters can serve as anchor points of differentiation trajectories. A combination of both types of information, however, is preferable. One could give up the detection of subpopulations and directly estimate the differentiation process from cell profiles. Clustering, however, implies a discretization that cannot capture the continuous nature of differentiation processes. Subpopulations are identified via clustering, yielding intuitive outcomes that can be validated by marker genes. Cell types can be characterized by expression profiles derived from single-cell RNA-seq.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |