This tutorial demonstrates how to analyze multi-section slides, focusing on exploring spatial and UMAP projections based on clusters and relabeling sections. We use one slice that contains thymus, small intestine, eyeball, whole brain, and lung as the example.
Please ensure that the tissue segmentation result meets the requirements of your downstream analysis.
Check and relabel
Some sections of the sample may be invalid due to detachment, folding, or necrosis. Please review your sample and relabel by selecting spots on demand.
Upload a CSV file (Optional)
If setting --split-tissue-into-labels in SAW count or SAW realign , you can find auto-labeling files in the pipeline output /outs/analysis/split_labels_*.csv.
Upload split_labels_*.csv to Custom groups.
The automatically generated labels are assigned consecutive numbers, which is very convenient for analyzing TMA samples. You can change the label name directly. And delete some labels with small spot numbers, which might be tissue fragments.
If there is no automatically labeled .csv file, try creating a new group and renaming it. Then, use lasso/rectangle selection to label the different parts.
Save the label successful Use eraser or brush tools to fine-tune the tissue border.
Save to a group after brushing Clustering and differential expression
Choose your target sections and perform clustering and differential expression again.
In this example, two thymus sections are selected for re-clustering. When clustering finishes, you can split the UMAP view by section and evaluate the batch effect.
After re-clustering, you can choose targeted clusters to perfrom differential expression analysis.
