`OpenEphysRawIO` header parse is slow

[weiglszonja]

Hey,

I have ecephys data from the Tye lab in legacy OpenEphys format (.continuous files) and I'm using OpenEphysRecordingInterface from neuroconv to write the data to NWB (which uses OpenEphysLegacyRecordingExtractor from spikeinterface).

I noticed that initializing the interface is slow, and I could trace it back to OpenEphysRawIO where
parsing the header for their recording takes at least 5 minutes. Do you have any idea what could contribute to this?

from neo.rawio import OpenEphysRawIO
# Parameters for conversion
neo_rawio = OpenEphysRawIO(folder_path)
neo_rawio.parse_header()

[JuliaSprenger]

@samuelgarcia Did you have a similar experience?

[samuelgarcia]

I did this long time ago. I remember that this format is nto optimal and so slow to parse/check.
I think mainly because of the spikes files but it is a guess.
Szonja could you check the speed when removing spikes files ?
If it is this we could add an option not not parse theses files.

[weiglszonja]

I think mainly because of the spikes files but it is a guess.
Szonja could you check the speed when removing spikes files ?

@samuelgarcia I don't actually have .spikes files in the recording folder that was shared with me.

[JuliaSprenger]

@samuelgarcia Any more ideas for quick checks on how to avoid this or should we start profiling?