This is abandoned. I don't remember the last time I used Seurat and have entirely moved to the scverse suite of tools for single-cell analysis.
R wrappers around dimensionality reduction methods found in Python modules. Uses the reticulate package to expose functionality. Additionally provides bridging functions that let these work as drop-in replacements when working with Seurat (verions 3) and SingleCellExperiment objects. Currently wraps:
- ForceAtlas2, using the Python implementation found here. Currently works for Seurat objects only (and will remain so until I can find a fast algorithm for computing simultaneous nearest neighbors for SingleCellExperiment objects).
- openTSNE
- UMAP (Uniform Manifold Approximation and Projection)
- PaCMAP (Pairwise Controlled Manifold Approximation)
- PAGA (Partition-Based Graph Abstraction) through Scanpy.
- PHATE (Potential of Heat-diffusion for Affinity-based Trajectory Embedding)
- PhenoGraph