AnnData Structure

AnnData is the internal object that Cellar uses for storing annotations and exporting session files. Here we describe the keys that Cellar uses to store its results.

---

For an overview of the AnnData parameters check the AnnData API.

• adata.X: The main data matrix

• adata.obs['labels']: Cluster IDs, integer
• adata.obs['annotations']: Cluster annotations (cell types), strings

• adata.obs['x'], adata.obs['y']: x, y coordinates corresponding to cell centers for spatial data

• adata.var['gene_symbols']: HGNC gene symbols, strings

• adata.uns['labels']: Clustering settings, dictionary
• adata.uns['x_emb']: Dimensionality reduction settings, dictionary
• adata.uns['x_emb_2d']: 2D dimensionality reduction settings, dictionary
• adata.uns['neighs']: Neighbors graph settings, dictionary

• adata.uns['spatial_idx']: 2D tile mapping each pixel to the corresponding cell (see spatial tile)

• adata.obsm['x_emb']: Reduced representation of the data, array
• adata.obsm['x_emb_2d']: 2D embeddings of the data, array
• adata.obsm['proteins']: Protein data for CITE-seq, array

• adata.obsm['genes']: Mapped gene expression data from CITE-seq for CODEX

• adata.obsp['neighs']: Neighbors adjacency matrix, sparse matrix

---