adataobs

adataobs refers to the observation metadata stored in an AnnData object, a data structure widely used for single-cell genomics analysis in Python with libraries such as Scanpy and anndata. The adata.obs attribute contains per-cell annotations in a tabular format, behaving like a pandas DataFrame. Each row represents an observation (cell) and each column a metadata attribute, with the row names aligned to adata.obs_names and to the corresponding rows of the expression matrix adata.X.

Commonly stored columns include biological and technical annotations such as sample, batch, donor, cell_type, and quality control metrics like percent_mito, n_counts, and n_genes. The exact columns depend on the dataset and analysis goals. This metadata is used to filter, group, or color cells in visualizations and to drive downstream analyses such as clustering, differential expression, and integration with other datasets. The adata.var attribute stores gene-level metadata, providing a parallel structure for features, while adata.obsm can contain per-cell embeddings produced by dimensionality reduction methods.

adata.obs is mutable and can be created or updated by assigning a DataFrame or new columns, with the requirement that the index matches adata.obs_names (the per-cell identifiers). Examples include adding a new column or subsetting by a condition, e.g., adata.obs['batch'] = batch_array or adata[adata.obs['cell_type'] == 'T_cell'].

In summary, adataobs is the per-cell metadata backbone of an AnnData object, enabling annotation, filtering, and integration of single-cell data alongside gene expression and dimensionality reduction information.