tsv output should not contain empty value for threshold and holding current

[jdcourcol]

it happens that the mecombo_emodel.tsv contains missing entries for threshold or holding current, leading to rows with 7 entries instead of 8 for instance.
This is incompatible with what the simulator expects (it expects 8 values to be provided).

Probably it relates to that line:

BluePyMM/bluepymm/select_combos/megate_output.py

Line 94 in 1f87b51

if null_rows.sum() > 0:

We may consider to discard the cells from the output for instance.

[arnaudon]

Did you get this warning message

BluePyMM/bluepymm/select_combos/megate_output.py

Line 100 in 1f87b51

print("WARNING ! There are rows with None for "

in your logs?

@wvangeit , shall we:

1. keep as it is
2. keep the warning and discard bad rows to get a clean output (in all files, not just extneurondb)
3. make MM crash as it is in the commented code?

I would be in favor of option 2, as jean-denis suggested.

[wvangeit]

As this shows, it looks like warnings should not be used, because nobody reads warnings.
Maybe the safest way is to make it crash by default, and have a flags like remove_rows_with_empty_threshold / remove_rows_with_empty_holding.
What do you think?

The only thing is, does it add an empty column, or no column at all? Maybe it would already be better to at least make sure an empty column is added?

[arnaudon]

Yes, this seems a good solution. Not sure which column you mean, w'll keep the same columns, but remove some rows, right?

[jdcourcol]

@arnaudon did you take a decision ?

[arnaudon]

I'll implement what werner proposed!

[arnaudon]

PR here: #225