generating results for a sequential modeling pipeline (to compare to our current pipeline) #132
Conversation
…eling (to compare to the current pipeline) for bootstrap_lassocv method. Note that a slight modification was made to the source code to utilize an additional keyword argument in generate_modeling_data, but this essentially doesn't change the source code at all
| # Rest of the code remains the same... | ||
| # Store the results under a new directory | ||
| sequential_output_dir = os.path.join( | ||
| output_dirpath, "sequential_top_genes_results" |
There was a problem hiding this comment.
in your results_20241202, I see this pattern:
results_20241202/ACE2/ACE2/
Where there is a directory with the same name nested under the other. As far as I know, this wasn't happening before. It isn't immediately clear to me that it is in this line that this occurs, but nowhere else stands out.
On a related note: somewhere in the docs, in the cmd line --help, or even as a README that gets deposited in every results otuput directory, there needs to be a definition of what "sequential_top_genes_results" are.
I agree with you that keeping this in the main workflow for now (meaning, not creating another cmd line option to turn this feature on and off) is a good idea. It just needs to be documented so that it is clear what these results are.
There was a problem hiding this comment.
Other than this, it looks good to me in terms of format. I haven't pulled and run this, and __main__ is where my testing is not strict. My strategy is to implement the logic elsewhere, and do only very simple things that I am quite certain of in __main__ (not a strong strategy). Please do make sure that you go over your added code carefully. Play with it in a console/notebook if you need to be sure of something.
modifying main.py to generate results for the sequential pipeline modeling (to compare to the current pipeline) for bootstrap_lassocv method. Note that a slight modification was made to the source code to utilize an additional keyword argument in generate_modeling_data, but this essentially doesn't change the source code at all