10x Genomics Support/Xenium Explorer/Tutorials/

Checking Xenium Data Quality

Data quality assessment is a sample- and experiment-specific process. Here are some of the ways in which Xenium Explorer can help to check data quality. Contact [email protected] for additional help.

These screenshots come from 10x public datasets. Click the links to download and explore these mouse brain and human lung cancer datasets.

Inspect nuclei-stained (DAPI) images for tissue detachment or tears using the image options and projection settings. For data generated with the cell segmentation staining workflow, the interior protein stain (alphaSMA/Vimentin) image may help to identify morphological structure in tissues such as connective tissue or blood vessels.

Xenium Explorer displays nucleus and cell segmentation polygon boundaries, which are approximations for visualization performance. This enables qualitative assessment of segmentation results based on either nuclear expansion or multimodal cell segmentation algorithms. The true segmentation mask results are in the cells.zarr.zip output file.

Display or export information about poorly segmented cells by clicking on a specific cell or using the lasso tools to select a region of interest.

To remove some cells from downstream analysis, export cell IDs from Xenium Explorer and filter them out from the cell-feature matrix.

For single channel (DAPI-only) datasets, examine cell segmentation by nuclear expansion results with a single color, such as white cell boundaries and red nucleus boundaries, to check cell segmentation quality.

For multimodal cell segmentation datasets, select Segmentation Method from the cell color dropdown in the Cells menu. Use the Image and Cells menu check boxes to compare the nucleus and cell boundaries with the DAPI, boundary, and interior stain images.

The segmented cell boundaries should corresond to the stain boundaries, except in the case of cells segmented via isotropic nuclear expansion. For nuclear expansion, the expansion stops when it encounters another boundary or if it reaches 5 µm (see segmentation algorithms).

Boundary stainDAPI + Interior RNA stains

The segmentation method per cell is shown when you click on individual cells in the Selections pop-up window (learn about multimodal cell segmentation algorithms). You can also open the analysis_summary.html in Xenium Explorer under Sample Information to view segmentation method metrics.

Examine cells colored by cluster affiliation to check whether clusters align with knowledge of the tissue type/state. Click on cells to check whether selected marker genes in cells match biological expectations.

Here are some ways to validate the accuracy of your transcript data with genes that have known spatial or cell localization patterns:

  • Select cell type marker genes and assess them using per-gene localization plots (Q-Score ≥ 20).
  • Select a group of marker genes for a given cell type and view as different colored points to visualize separation between cells (i.e., to identify cell subtypes).
  • View gene transcript patterns individually in the viewing area.
  • View all gene transcripts as a density map across the entire sample to ensure uniform result.