Single cell azimuth projection

Single-Cell Azimuth Projection

Single-cell sequencing provides an incredibly detailed description of the genes being expressed by individual cells from a complex tissue. However, it can sometimes be difficult to interpret this information in a coordinated way across multiple specimens, analysis batches, or sampling modalities. The Satija lab published a reference-based approach (Azimuth), which projects newly collected datasets into the multidimensional space established for a set of curated tissue-specific atlases. This approach can be used to quickly annotate cell types and align UMAP ordinations for new datasets to facilitate rapid comparison.

The Azimuth analysis can be run on single-cell gene expression datasets in Cirro, and will produce as an output an updated Seurat (h5seurat) or Scanpy (h5ad) object which can be used for further downstream analysis.

Azimuth Human Motor Cortex Reference Atlas:

Azimuth References:

• Human Adipose - adiposeref Cells in Reference: 160,075
• Human Bone Marrow bonemarrowref Cells in Reference: 297,627
• Human Fetal Development - fetusref Cells in Reference: 377,456
• Human Heart - heartref Cells in Reference: 656,509
• Human Kidney - kidneyref Cells in Reference: 64,693
• Human Lung v2 - lungref Cells in Reference: 584,944
• Human Motor Cortex - humancortexref Nuclei in Reference: 76,533
• Mouse Motor Cortex - mousecortexref Nuclei in Reference: 159,738
• Human Pancreas - pancreasref Cells in Reference: 35,289
• Human PBMC - pbmcref Cells in Reference: 161,764
• Human Tonsil - tonsilref Cells in Reference: 263,299

Citation:

• Hao, Yuhan et al. “Integrated analysis of multimodal single-cell data.” Cell vol. 184,13 (2021): 3573-3587.e29. doi:10.1016/j.cell.2021.04.048