Entering edit mode
8 weeks ago
Saeedeh Salehi
▴
20
I have a single-cell RNA-seq dataset that includes:
16 PBMC samples from tumor patients and 4 healthy individuals 16 Tumor tissue samples from the same tumor patients 3 normal tissue samples adjacent to the tumors
I want to compare gene expression in T cells from PBMCs with T cells from tumor tissue to identify candidates genes that suppressed in the TME. However, I am concerned about the biological and technical differences between PBMCs and tumor tissue.
Would this be a valid comparison? What factors should I consider (e.g., batch effects, microenvironmental influences, cell state differences) to ensure meaningful results?
Any insights or suggestions on the best practices for such an analysis would be greatly appreciated!
How was the tumor digested?
Thank you for your response! I am using the GSE155698 dataset. According to the authors, tumor tissues were mechanically minced and enzymatically digested with collagenase P (1 mg/mL DMEM), followed by filtration through a 40micrometer mesh to obtain single cells. Dead cells were then removed using the MACS® Dead Cell Removal Kit.
Let me know if you have any thoughts on how this digestion process might influence the comparison between PBMC-derived and tumor-infiltrating T cells!
PBMCs are blood, so they're not digested. Tissue is digested, so incubated at 37°C with aggressive enzymes etc that can activate cells. Hence, comparing blood and tissue is inherently difficult. That is the bias to be expected. Meaning, more inflammation or activation in in the tissue group can well be a processing artifact. Whether this is true for your data I cannot tell, but it's the major concern of any tissue vs blood comparison. On top of the technical differences there is also the stroma effect so extravasated cells per se could have different transcription that circulating cells. The Q is whether a blood cell is a good proxy for a cell in a tissue. That is on you to decide for your biological system.