vitkl / cell2location_paper Goto Github PK

Hello @vitkl ,

Thank you for writing cell2location and releasing the notebooks replicating your results. I have been trying to replicate the synthetic benchmark pipeline to replicate the results and play around with the cell2location model, but have been running into some issues navigating the different notebooks and getting the data required to run them.

From reading the repo history, the code and following the issue tickets from this repo and cell2location, I have gathered that the relevant notebooks are:

• cell2location_paper/notebooks/benchmarking/synthetic_data_construction_improved_real_mg.ipynb : for the generation of the synthetic data, rather than the other two notebooks.
• cell2location_paper/notebooks/benchmarking/Fig1BCDE_FigS1S2S3_synthetic_data_improved_revision4_plots.ipynb for the loading and comparison of the results obtained from the different trained models (cell2location, spotlight, etc.)

My primary aim is to use instantiate my own cell2location model, train it, run it against the synthetic data, and obtain some quantitative results and do it again with some other hyperparameters, etc. Hence I am essentially struggling to find the notebook that lies inbetween the two logical steps above. Could you guide me on how I could do this?

Best regards,
paulmorio

Hi,
I want to run the simulation that was used in the paper.
I am walking through the jupyter notebook, but I can't find all of the folders/files in the repository:

• /cell2location_paper/notebooks/selected_data/mouse_visium_snrna/
• sp_data_file = f'{c2l_results_folder}{run_name}/sp_with_clusters.h5ad'
• adata_snrna_raw = anndata.read(f'{sc_data_folder}rawdata/all_cells_20200625.h5ad')

Help would be appriciated!