Single-cell mapping of gene expression landscapes and lineage in the zebrafish embryo
估计阅读时长: 7 分钟

Assembles a manifold that is defined through a series of overlapping, locally-defined PCA subspaces.

  1. Non-mutual k-nearest-neighborhoods are first obtained for each cell in timepoint i. Neighbor edges are queried from timepoints i (within-timepoint edges) and i-1 (link edges) after projecting into a PCA subspace defined by all cells from timepoint i.
  2. Outgoing edges are then subject to local and global neighborhood restictions.
  3. The graph is restricted to mutual edges.

Fig. 2. Single-cell graph reveals a continuous developmental landscape of cell states. (A) Overview of graph construction strategy, and a force-directed layout of the resulting single-cell graph (nodes colored by collection timepoint). For each cell, up to 20 within- or between-timepoint mutual nearest neighbor edges are retained. (B) Single-cell graph, colored by germ layer identities inferred from differentially expressed marker genes (see table S2). (C) Single-cell graphs, colored by log10 expression counts for indicated cell type-specific marker genes.

A single-cell graph of cell state progression in the developing zebrafish embryo

We sought to map trajectories of cell state during develop-ment by linking cell states across time. Several computational approaches exist to infer orderings of asynchronous pro-cesses from scRNA-seq data (9–11), typically by projecting all cells into a single low-dimensional latent space. Such strategies may be illsuited to map gene expression in developing embryos, which exhibit dramatically increasing cell state di-mensionality and continuous changes in the sets and num-bers of cell state-defining genes (fig. S2, D and E).

To overcome these obstacles, we developed a graph-based strategy for locally embedding consecutive timepoints on the basis of biological variation that they share, rather than using a global coordinate system for all timepoints.

  1. This approach first constructs a single-cell k-nearest-neighbor graph for each timepoint ti, with nodes representing cells and edges linking neighbors in a low-dimensional subspace;
  2. it then joins the graphs by identifying neighboring cells in pairs of adjacent time points, using a coordinate system learned from the future (ti+1) timepoint (see methods).
  3. The resulting graph spans all time points, and allows application of formal graph-based methods for data analysis.

Single-cell mapping of gene expression landscapes and lineage in the zebrafish embryo. Wagner DE, Weinreb C, Collins ZM, Briggs JA, Megason SG, Klein AM. Science 26 Apr 2018. doi:10.1126/science.aar4362

高级数据科学家 at 苏州帕诺米克
Working on Engineered bacteria CAD design on its genome from scratch. Writing scientific computing software for Tianhe & Sunway TaihuLight supercomputer. Do scientific computing programming in R/R# language, he is also the programming language designer of the R# language on the .NET runtime.
Latest posts by 谢桂纲 (see all)

Order by Date Name

One response

Leave a Reply

Your email address will not be published. Required fields are marked *

July 2024
  1. 针对图对象进行向量化表示嵌入: 首先,通过node2vec方法,将node表示为向量 第二步,针对node向量矩阵,进行umap降维计算,对node进行排序,生成node排序序列 第三步,针对node排序序列进行SGT序列图嵌入,实现将网络图对象嵌入为一维向量