Description:This plugin allows reading structural properties of AnnData h5ad files in nextflow. Such files are mostly used in single-cell transcriptomics to store sparse count matrices with additional metadata.
As the file format is based on the h5 format, it is not necessary to load the entire file into memory, but only parts that are relevant. For example, one can check if a certain layer or obs column is present, without loading the entire count matrix into memory. This is useful in nextflow, because we can implement branching, filtering, and pipeline input validation without having a dedicated process.
This plugin is essentially a wrapper around the jhdf package. It provides an interface to the anndata properties that closely resembles the Python implementation, which is what most users will be familiar with.
The plugin is limited to reading structural properties and metadata. It does not allow accessing the contents of the count matrices and obsm/varm fields, as this would require a large amount of memory and should thus be done in a dedicated nextflow process.
License: MIT