[Mne_analysis] old versus new event files

[A. Klimovich-Smith]

Hello everyone,

I have a question about the 'old' versus 'new' event (.eve) file formats 
and how those are treated by the mne_process_raw function during 
epoching.

In one of my single-trial analyses I had to replace trigger values in 
the raw event files with custom values. I ran those and all seemed well. 
Now I am re-using the same .eve files again and noticed when they are 
read with the mne_process_raw I get a warning that these event files are 
detected as 'old format'. This is happening because when replacing the 
triggers I also replaced the trigger value in first line of the event 
file (which is meant to be 0) to some non-sense value (7777 - as for all 
events of no interest). The first line of the event file encodes the 
first pseudo event and the timing in this line and the sample number are 
critical so if I replace these, all subsequent epochs will be 
misaligned. But I didn't change these - just the value of the digital 
trigger channel (i.e. ,the <to> value).

MNE manual (I am using 2.7.3 version), however, warns not to touch this 
first line at all. As this line helps the software to interpret the file 
format ('old' vs 'new'). But I can't find anywhere what this actually 
means and how would the epochs be treated in the old format? I only know 
that in the old format .eve file did't have this first empty event. But 
new mne is backward compatible and deals with this somehow.

So I ran with both 'old' (with 7777 non-sense trigger in the 1st line) 
and 'new' (the same but with correct 0 trigger value) event files, 
looked at the single trial epochs and they are identical (I get a matrix 
of 0s when i subtract one from the other). So the question is - does the 
trigger value of the first line matter then? Or would both 'old' and 
'new' versions of the event file be treated in the same way - i.e the 
value of the first sample in the first line be subtracted from the 
subsequent epochs?


Thanks,
Ana