Fast single-cell analysis with non-negative dimensional reductions
Details
There are reasons to not use PCA.
PCA fits to missing signal,
considers only highly variable features,
is almost useless without further graph-based analysis,
requires centering and scaling of your data,
and is robust only within experiments.
Instead, you should use Non-negative Matrix Factorization (NMF).
NMF imputes missing signal,
learns models using all features,
does everything PCA does and provides useful information itself,
requires only variance stabilization,
and is robust across experiments.
Singlet is all about extremely fast NMF for single-cell dimensional reduction and integration.
See the vignettes to get started.