Overview of the Cerebro v1.3 class

The introduction of the Cerebro_v1.3 class is an attempt to formalize the data structure and thereby provide stability in future releases, ultimately aiming to improve backwards compatibility between Cerebro version.

While the data is organized similar to how it was done before, now they live in their own class with setter and getter methods for the different slots. Also, unnecessary prefixes and suffixes have been removed, allowing direct connection between results, e.g. tables of marker genes, and the grouping variable they belong to / were derived from. Moreover, additional layers have been added to allow storing results derived from different methods.

Let’s look at some details. Overall, the object is organized as such:

<Cerebro_object>
├── version (package_version)
├── experiment (named list)
│   ├── experiment_name (string)
│   ├── organism (string)
│   ├── date_of_analysis (string)
│   └── date_of_export (string)
├── technical_info (named list)
├── parameters (named list)
├── groups (named list)
│   └── <grouping_variable> (vector of group levels)
├── cell_cycle (vector)
├── gene_lists (named list)
│   └── <gene_list_name> (vector of genes)
├── expression (matrix-like)
├── meta_data (data.frame)
├── projections (named list)
│   └── <projection_name> (data.frame)
├── trees (named list)
│   └── <grouping_variable> (phylo)
├── most_expressed_genes (named list)
│   └── <grouping_variable> (data.frame)
├── marker_genes (named list)
│   └── <method> (named list)
│       └── <table_name> (data.frame)
├── enriched_pathways (named list)
│   └── <method> (named list)
│       └── <table_name> (data.frame)
├── trajectories (named list)
│   └── <method> (named list)
│       └── <trajectory_name> (depends on method, most likely a list or data frame)
└── extra_material (named list)
    ├── tables (named list)
    │   └── <name> (data.frame)    
    └── plots (named list)
        └── <name> (ggplot-object)

Interaction with the Cerebro object will be done through internal methods, providing additional control for the correct format of the object.

As long as you follow the described scheme, you can export and visualize any data frame that is stored in the marker_genes or enriched_pathways slots. I have prepared a vignette with some more details that you can find here.

Please make sure to use names for items in lists, otherwise it will likely result in an error.

Information stored in the technical_info and parameters slots will be presented in the “Analysis info” tab. Here you can add strings, numbers, and lists that with you think are relevant to share.

Calling the Cerebro object will give an overview of the information it currently contains:

class: Cerebro_v1.3
cerebroApp version: 1.3.0
experiment name: pbmc_Seurat
organism: hg
date of analysis: 2020-02-19
date of export: 2020-09-08
number of cells: 5,697
number of genes: 15,907
grouping variables (3): sample, seurat_clusters, cell_type_singler_blueprintencode_main
cell cycle variables (1): cell_cycle_seurat
projections (2): UMAP, UMAP_3D
trees (3): sample, seurat_clusters, cell_type_singler_blueprintencode_main
most expressed genes: sample, seurat_clusters, cell_type_singler_blueprintencode_main
marker genes:
  - cerebro_seurat (3): sample, seurat_clusters, cell_type_singler_blueprintencode_main
enriched pathways:
  - cerebro_seurat_enrichr (3): sample, seurat_clusters, cell_type_singler_blueprintencode_main, 
  - cerebro_GSVA (3): sample, seurat_clusters, cell_type_singler_blueprintencode_main
trajectories:
  - monocle2 (1): highly_variable_genes

For more details, please check the Cerebro_v1.3 reference page.

Roman Hillje

12 March, 2021