Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1.48 KB

v1_0.md

File metadata and controls

35 lines (22 loc) · 1.48 KB

Overview

Contents are stored inside an inputs HDF5 group at the root of the file. The cell_labelling group itself contains the parameters and results subgroups.

Definitions:

  • num_clusters: number of clusters generated in the analysis. This is determined by the choose_clustering step.

Parameters

parameters should contain:

  • human_references: a string dataset defining the human reference datasets used for labelling. Each entry contains the name of a reference dataset, e.g., "BlueprintEncode".
  • mouse_references: a string dataset defining the mouse reference datasets used for labelling. Each entry contains the name of a reference dataset, e.g., "ImmGen".

Results

results should contain:

  • per_reference: a group containing the label assignments for each cluster in each reference. Each child is named after its corresponding reference as listed in parameters, though not all references listed in parameters need to be present here. Each child is a 1-dimensional string dataset of length equal to num_clusters, where each entry contains the assigned label for the corresponding cluster.

For multiple references of the relevant species, results will also contain:

  • integrated: a string dataset of length equal to num_clusters. This specifies the reference with the top-scoring label for each cluster, after integrating the results of all per-reference classifications.

History

Added in version 1.0.