PBMC analysis

[wes-lewis]

Can you make public the code for your PBMC analysis? Or uploading the adata object you use? Trying to run the same code as the tutorial, but using the 3k PBMC dataset from scanpy, is currently causing the code to stall on my system. I've tried downsampling the pbmcs to ~ 400 cells and 1000 genes, ensuring the same columns exist in the .obs for pbmc's as for your example dataset, etc. Mentioning expected runtimes for the PBMC dataset would also be helpful. I'm running on a compute cluster with GPU support, and see no reason for the issues I'm having.

All the best
Wes

[chanwkimlab]

My apologies for my belated response. I missed this issue.

The analysis code for PBMC is available here: https://github.com/chanwkimlab/MarcoPolo/blob/master/experiments/
The code includes the entire pipeline from downloading, preprocessing to running MarcoPolo.
Please note that the code is in R. The MarcoPolo software was originally implemented in R but has been reimplemented in Python to handle the package compatibility issues better.

I'd glad to help you debug the issue. Were you able to run the code in the tutorial using the example data on your machine? Or did you encounter any error messages? For reference, the analysis of PBMC dataset was completed in several hours on our machine.