File not found

[jonesse3]

When trying to run cell_autoencoder.py, I got an error saying that a file was not found.

How can I get the following files in order to reproduce the model:

• cell_index_file = 'data/cell/cell_30_ge_cn_id.csv'

• cell_ge_file = 'data/cell/ONeil_31_cell_47838_dim_genex.csv'

• cell_cn_file = 'data/cell/ONeil_31_cell_23316_dim_CNV.csv'

• cell_ge_ae = "data\cell\ONeil_30cell_ge_512dim.pt"

• cell_cn_ae = "data\cell\ONeil_30cell_cn_512dim.pt"

[yuupang]

These files were processed. Original file can be downloaded by CCLE. The files can be generated by the paper.

[jonesse3]

Can I get the processed files so I can reproduce the results from the paper? Otherwise I can't run the code per the instructions in the README.

I went to the links provided, but it's unclear what files and their versions I should download from depmap. Once these are identified, do I run preprocess.py to process these files and generate the autoencoders (also missing from the data/cell folder)?

[cindyhuan]

Thank you for your reply and clarification.

I understand that the original data is from CCLE and that the ONeil_... files are processed versions based on the methods described in your paper.

However, trying to reproduce the exact processed files (like ONeil_31_cell_47838_dim_genex.csv and ONeil_31_cell_23316_dim_CNV.csv) just from the high-level description in the paper is extremely difficult. Any small differences in data filtering, normalization, or cell line selection will result in files that do not match what is required by the provided scripts (like cell_autoencoder.py).

To ensure the reproducibility of your valuable work, would it be possible for you to please share the data processing scripts that you used to generate these specific ONeil_... and cell_30_... files from the original CCLE data?

Alternatively, if sharing the scripts is not feasible, would you consider uploading the final processed data files to a repository (like Zenodo, Figshare, or even a Google Drive link)?

Thank you for your help in making this research reproducible.

Best regards