search

OSU-BMBL / scDEAL

Deep Transfer Learning of Drug Sensitivity by Integrating Bulk and Single-cell RNA-seq data
Apache License 2.0
46 stars 11 forks source link

no results generated when running my own dataset #13

Open watermelontreesjs opened 1 year ago

watermelontreesjs commented 1 year ago

Hello, scDEAL is a very useful tool. I'd like to run it on my own dataset for drug prediction. Here's my code in linux bash:

python bulkmodel.py --drug "I.BET.762" --data "/data/scDEAL/data/ALL_expression.csv" --label "/data/scDEAL/data/ALL_label_binary_wf.csv" --device gpu

python scmodel.py --sc_data "/data/scDEAL/data/own.csv" --drug "I.BET.762" --checkpoint "False" --mod ori --bulk_data "/data/scDEAL/data/ALL_expression.csv" --label "/data/scDEAL/data/ALL_label_binary_wf.csv"

own.csv is my own single cell dataset used for drug prediction, the bulkmodel.py step generated a file named "integrate_data_GSE110894_drug_I.BET.762_bottle_32_edim_512,256_pdim_16,8_model_AE_dropout_0.3_gene_F_lr_0.01_mod_new_sam_no" in the "scDEAL/save/bulk_encoder/" directory.

but when i continue to run the scmodel.py, the step stop without warning or error, and no results were generated, only some printing in my terminal

save/bulk_pre/integrate_data_/data/scDEAL/data/own.csv_drug_I.BET.762_bottle_512_edim_512,256_pdim_16,8_model_AE_dropout_0.3_gene_F_lr_0.01_mod_ori_sam_no

(507, 26609)

two databases combine

How can i solve this problem, could you provide more detailed advice on how to use your own bulk dataset and single-cell dataset for the scDEAL process?

juychen commented 1 year ago

Hi, the error message may be stored in the folder save/logs. Could you please check it？

valerychu commented 9 months ago

I managed to run on my own data by modifying the code in scmodel.py

DATA_MAP={ "GSE117872":"data/GSE117872/GSE117872_good_Data_TPM.txt", "GSE110894":"data/GSE110894/GSE110894.csv", "GSE112274":"data/GSE112274/GSE112274_cell_gene_FPKM.csv", "GSE140440":"data/GSE140440/GSE140440.csv", "GSE149383":"data/GSE149383/erl_total_data_2K.csv", "GSE110894_small":"data/GSE110894/GSE110894_small.h5ad" }

You can add the mapping of your own data like:

"GSE111111_my":"data/GSE111111/GSE111111_my.csv"

and in the parameter use:

python scmodel.py --sc_data GSE111111_my

seekning commented 4 months ago

do you have wechat or QQ/?i want to face to face comuunicate with you about your code,it makes me so confused