Identifying tumor vs normal cells in scRNAseq samples
1
2
Entering edit mode
20 months ago
bompipi95 ▴ 150

Hi everyone,

I currently have the gene expression matrix for mouse scRNA-seq bladder samples in WT vs KO conditions. These samples contain a mixture of normal (immune & bladder cells) and tumor cells. I have carried out the standard quality-control, normalisation, dimensional reduction and integration steps with Seurat v4.0.6. My question here is: how can I identify the tumor vs normal cells in the scRNA-seq dataset apart from one another?

I currently have 2 ideas:

  1. Obtain the marker genes for the clusters and then have experts identify which clusters correspond to normal cells and which are likely the tumors. I can also perform automated cell-type annotation (e.g. reference-based annotation) against age-matched reference mouse bladder samples, which will help to narrow down the clusters that correspond to normal tissue samples, leaving the remainder as likely candidates for tumor-cell containing clusters.

  2. Identify tumor cells as cells that have CNV. Identify CNVs using tools like honeybadger & inferCNV, and based on that, probabilistically label cells as tumor / normal.

I would appreciate your suggestions on the matter, as well as references to papers that have performed similar analysis. Thank you for reading!

seurat CNV scRNA-seq cancer • 918 views
ADD COMMENT
1
Entering edit mode

You could also check the mitochondrial variants to potentially identify the cancer clone:

http://www.bioconductor.org/packages/release/bioc/html/mitoClone2.html

ADD REPLY
3
Entering edit mode
20 months ago
Rafael Soler ★ 1.2k

Hello,

You can do both. Check for common marker genes first, and then test cells for CNV. I would also recommend that you check out the latest tools for CNV recognition, as Identifying tumor cells at the single-cell level using machine learning

ADD COMMENT

Login before adding your answer.

Traffic: 2498 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6