Regarding Testing on new data

[SmritiChawla]

We are trying to predict drug response on a new dataset i.e. cell lines present in the new dataset is not present in the training dataset. However, I am getting the following error as shown in the screenshot below;

How can we predict on a new set of cell lines?

[rafaelperes]

Hi @SmritiChawla! Could you give us more details? Are you using the notebook?

Best,
Rafael.

[sLee0709]

Hi @SmritiChawla! Could you give us more details? Are you using the notebook?

Best, Rafael.

Hi Rafael, I'm facing the same error. I'm firstly using model_training.py to train models based on GDSC data, and then use the trained models to make predictions towards our own single cell expression matrix. In the line 60 of model.py, np.matrix(test_kernel_df[kernel_sample_list]) failed, as it seems you used the cell line IDs of GDSC data as indexes for test_kernel_df. But the column names of test_kernel_of are cell label ids generated from scRNA sequencing. It is impossible to do that unless you used the same data. I'm not sure how did you do when you used your method towards new scRNA data?