How to perform perturbation prediction with 10X tissue dataset?

[Junjie-Hu]

Dear team, Thanks a lot for providing scGPT for biological researchers. You provide a tutorial for perturbation prediction using Adamson or Norman's datasets. However, I am still confused about how to do it in my dataset (10X lung cancer). Adamson or Norman's datasets are based on Perturb-seq platform. I just want to predict what genes will change if I knock out a gene (e.g. TCF7) in T cells in my datatset. Can I get it using scGPT? Could you please provide a tutorial for this?

Thank you.

[subercui]

Hi, thank you for the interesting question! Could you provide more details about your data, do you mean you have scRNA-seq data of T cells in lung cancer tissue samples? The data is rather observational only?

[Junjie-Hu]

Thanks for your response, Indeed, I want to predict gene expression changes following the in silico knockout of TCF7 in CD8+ T cells from lung cancer tissue. Could you kindly provide guidance on how to approach this analysis? I would greatly appreciate it if you could offer a tutorial or resources for analyzing scRNA-seq data using the Peripheral Blood Mononuclear Cells (PBMC) dataset, which is freely available from 10X Genomics. The raw data can be accessed using the following link: https://cf.10xgenomics.com/samples/cell/pbmc3k/pbmc3k_filtered_gene_bc_matrices.tar.gz

[jackbrougher]

Similarly super curious about performing this type of assessment. We have several different observational datasets of rare tissue from otherwise healthy individuals - is it possible to model gene KO pertubations?