Cells & embeddings

With hundreds to thousands of cells, the question becomes which cells are alike. uchrom.emb embeds cells from any per-cell matrix — RNA counts, chromatin marks (IF), accessibility (ATAC), contact maps — clusters them, finds marker features, and stores the result in cd.cellm with the metadata the web browser uses to label it.

Task	API
embed cells from RNA / IF / ATAC / any matrix (PCA, t-SNE, UMAP)	uchrom.emb.embed_cells, normalize_counts, tfidf_lsi
cell features from contact maps	scool_cell_features, schicluster_features
clusters and their markers	cluster_cells, marker_features
per-cell averages of spot / bin signals	aggregate_tracks
Higashi / FastHigashi embeddings of single-cell Hi-C	uchrom.emb.higashi (extra emb)

Tutorials

• Cell embeddings and clusters across modalities
• Contact-map cell embedding with FastHigashi

API: uchrom.emb.