A repo to capture cell embedding visualizations
In the handbook folder, a Visualization Handbook is used to justify our visualization settings and showcase how our design evolved.
- Basic Visualization - Mapping the cell type labels with the h5ad source file: here, we adopt two kinds of normalization strategies to normalize the matrix in h5ad source file to improve the readability of the visualization. As shown below:
- Visualization Tuning - PCA-200 + tSNE / Visualization Tuning - PCA-1000 + tSNE: we adopt the PCA algorithm for 200/1000 dimensions and then for t-SNE algorithm, with the same normalization settings in the Basic Visualization jupyter notebook. The visualization for the PCA setting of 1000 is shown below:
- Visualization Tuning the top 24 cell types - PCA-200 + tSNE: we visualize the top 24 cell types as shown below:
