Sparse raw counts and UMAP

JonathanShor / DoubletDetection

Doublet detection in single-cell RNA-seq data.

https://doubletdetection.readthedocs.io/en/stable/

MIT License

81 stars 23 forks source link

Sparse raw counts and UMAP #118

Closed adamgayoso closed 5 years ago

adamgayoso commented 5 years ago

Fixes #117
Reduces memory overhead of storing both normed counts and raw counts as dense
Reduces per iteration speed by about 3 seconds on laptop
Exchanges TSNE with UMAP for visualization (much faster)
If a dense raw_counts is entered into .fit() it will automatically be sparsified

adamgayoso commented 5 years ago

I am going to push an updated notebook, but we also need to change normalizer to take a sparse input (in the docstring). That is to say any custom normalizer must accept a sparse array.