Issues with Regtools Statistical Analysis Workflow

[mguaita]

Dear Griffith Lab,

Thank you very much for this tool.

Following the Example workflow (https://regtools.readthedocs.io/en/latest/workflow/), I have encountered some issues during the statistical analysis of the Regtools comparison outputs.

1. Does the compare_junctions_hist.py script work properly with the recommended option variant_grouping_mode == 'exclude'? I get all zeros for the non-mutated scores.
   In lines 235-236, all the rows with common junctions between the samples_w_variant_df are removed from the samples_wout_variant_df, thus removing the score values of the non-mutated samples for the junctions of interest. This makes sense for variants associated with the same junction, but later the script seems to not restore the non-mutated scores well, generating a list of all zeros for the non-mutated scores. I do get p-values with variant_grouping_mode == "strict".

2. Is the R script compare_junctions_hist_v2.R a good alternative for the statistical analysis?

I would really appreciate some assistance with this final step, as the rest of the tool works well.