Open PierreLaville opened 3 months ago
Furthermore, in the example above the thresh parameter (0.8
) uses the .
character which could lead to some issues (confusion with file extension, harder parsing with regular expressions, etc.), so should be avoided.
Indeed, we can go on two different way:
keep the parameters in the file name. In this cases there should be a function that is taking care of preparing the parameters string to plug into the filename
use a log file with the same extension as the output file (or main directory) and with some lines listing all the parameters.
I prefer when the file creations are explicitly asked by the user, so I would vote in favor of solution 2. (with solution 1. the file creation is implicit as a function of the run parameters).
The problem with the float parameter 0.8
for the threshold also votes in favor of solution
As discussed, option 2. also enhances reproducibility (keeping track of which command has been run on which file). See Logging module in Python.
Improve naming consistency for parameters in output files:
test_pair_test_run_pep_df_mindp_20_maxdp_400_minad_5_gq_20_thresh_0.8_bl_3.pkl
test_pair_test_run_mismatches_20_400_5_gq_20_0.8_bl_3.tsv
Or log parameters in a specific file.