This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Is it expected for T cells to be broadly distributed across the UMAP in pancreatic cancer scRNA-seq data?

I processed pancreatic cancer single-cell RNA-seq data using Scanpy and performed cell type annotation with CellTypist.

After annotation, I noticed that T cells are broadly distributed across the UMAP space rather than forming one or a few compact clusters.

I am wondering whether this pattern is biologically expected or if it could indicate potential issues with annotation, batch effects, or dimensionality reduction.

The dataset includes tumor tissue (and liver samples in some cases), and multiple T-cell subtypes (e.g., effector, memory, innate-like T cells) are present. Canonical T-cell markers (CD3D/E, TRAC) are clearly expressed, and NK cells and other immune lineages appear well separated.

In this context, is a widespread distribution of T cells on UMAP a normal observation reflecting T-cell heterogeneity, or should T cells typically form a more compact region in the embedding? enter image description here

enter image description here

enter image description here

enter image description here

enter image description here

annotation single_cell_rna_seq umap scrna-seq
• 923 views • link
updated by yura.grabovska • yura.grabovska • written by Bum • Bum

1 answer

You should specifically check CD4 and CD8A/B. I have definitely previously seen data like 10x scRNASeq showing slight CD45+ signal in populations where I wouldn't expect it. It looks like the epithelial cells are lighting up but are generally lower than the cluster where the T cells appear to be. It would be worth looking at each individual gene in your T-cell signature and see if a handful are biasing the expression and whether they are truly T-cell unique or if they are expressed on the surface of your cancer sometimes.

• 0 views • link
written by yura.grabovska • yura.grabovska

Log in to answer this question.

More posts like this