celltype-deconvolution

Spatial transcriptomics deconvolution using RCTD infers per-spot cell-type proportions by mapping Visium-like spot data against an annotated single-cell reference, producing dominant celltype labels stored in adata.obs.

Reference-based deconvolution requires raw UMI counts in spatial data, an annotated single-cell reference with celltype labels, and overlapping gene sets between the spatial and reference datasets to execute successfully.

Yes, scRNA-seq data serves as the annotated single-cell reference for RCTD deconvolution, estimating cell-type proportions for each spatial feature and assigning dominant cell types to Visium-like spots.

RCTD estimates cell-type proportions per spatial spot using a single-cell reference, then derives a dominant cell type for each spot, storing the results in adata.obsm['deconv_weights'] and adata.obs['celltype'].

Yes, RCTD deconvolution is applicable for annotating tumor microenvironments on Visium slides by mapping spatial spots to immune and tumor cell types using an annotated single-cell reference.

RCTD deconvolution requires overlapping gene sets between spatial and reference data, raw UMI counts, and an annotated single-cell reference; spots lacking these prerequisites cannot be accurately deconvolved.