Introduction

The in-memory module allows users to load the complete anndata object into memory. This is useful for small to medium-sized datasets, where the entire dataset can be loaded into memory with overhead for the memory requirements of the methods (although we aim that this stays minimal). For now the in-memory module is the only module that allows for transformations of the data (e.g. normalization and log-transformation).

The in-memory module is the module we're working currently on the most, as it is for now the most important module to build high-performance single-cell analysis pipelines, as they stay typically relatively small.