Question: How to use Step 1 .pt output file as --gene_exp input in Step 2?

[strMinku]

Hello,

I’ve been following the provided example to build a model for scPrediXcan using step 1 and step 2.

In step 1, I used the example file Step1_example_input_expression.csv as input, and the output I obtained was a .pt file.

From my understanding, the output from step 1 should be used in step 2 as the --gene_exp argument. However, the file generated in step 1 is a .pt file, and I’m not sure if it can be directly used, or if it needs to be converted into a different format before being passed to --gene_exp.

Could you clarify whether an additional conversion step is required? If so, could you provide guidance on how to perform that conversion?

Thank you!

[Yichao-Zhou]

Hi strMinku,

Thanks for reaching out!
The .pt file from the step 1 is just the file storing the ctPred model weights. Then with the personalized epigenomic features as the input, you can use the ctPred model (with .pt file) to predict the personalized expressions and make them into a gene-by-individual file. That is the file put into the --gene_exp at step2.
If you have any more questions, feel free to reach out to me @yichaozhou@uchicago.edu.

Best,
Charles